0
Diagnosis and Management of Lung Cancer: ACCP Guidelines (2nd Edition) |

Noninvasive Staging of Non-small Cell Lung Cancer*: ACCP Evidenced-Based Clinical Practice Guidelines (2nd Edition) FREE TO VIEW

Gerard A. Silvestri, MD, FCCP; Michael K. Gould, MD, MS, FCCP; Mitchell L. Margolis, MD, FCCP; Lynn T. Tanoue, MD, FCCP; Douglas McCrory, MD; Eric Toloza, MD, FCCP; Frank Detterbeck, MD, FCCP
Author and Funding Information

*From the Department of Medicine (Dr. Silvestri), Medical University of South Carolina, Charleston, SC; the Department of Medicine (Dr. Gould), Veterans Affairs Palo Alto Health Care System, Palo Alto, CA; the Department of Medicine (Dr. Margolis), University of Pennsylvania, Philadelphia, PA; the Departments of Medicine (Dr. Tanoue) and Surgery (Dr. Detterbeck), Yale University, New Haven, CT; and the Departments of Medicine (Dr. McCrory) and Surgery (Dr. Toloza), Duke University Medical Center, Durham, NC.

Correspondence to: Gerard A. Silvestri, MD, FCCP, Professor of Medicine, Medical University of South Carolina, Department of Pulmonary and Critical Care Medicine, 171 Ashley Ave, Room 812-CSB, Charleston, SC 29425-2220; e-mail: silvestri@musc.edu



Chest. 2007;132(3_suppl):178S-201S. doi:10.1378/chest.07-1360
Text Size: A A A
Published online

Background: Correctly staging lung cancer is important because the treatment options and the prognosis differ significantly by stage. Several noninvasive imaging studies including chest CT scanning and positron emission tomography (PET) scanning are available. Understanding the test characteristics of these noninvasive staging studies is critical to decision making.

Methods: Test characteristics for the noninvasive staging studies were updated from the first iteration of the lung cancer guidelines using systematic searches of the MEDLINE, HealthStar, and Cochrane Library databases up to May 2006, including selected metaanalyses, practice guidelines, and reviews. Study designs and results are summarized in evidence tables.

Results: The pooled sensitivity and specificity of CT scanning for identifying mediastinal lymph node metastasis were 51% (95% confidence interval [CI], 47 to 54%) and 85% (95% CI, 84 to 88%), respectively, confirming that CT scanning has limited ability either to rule in or exclude mediastinal metastasis. For PET scanning, the pooled estimates of sensitivity and specificity for identifying mediastinal metastasis were 74% (95% CI, 69 to 79%) and 85% (95% CI, 82 to 88%), respectively. These findings demonstrate that PET scanning is more accurate than CT scanning. If the clinical evaluation in search of metastatic disease is negative, the likelihood of finding metastasis is low.

Conclusions: CT scanning of the chest is useful in providing anatomic detail, but the accuracy of chest CT scanning in differentiating benign from malignant lymph nodes in the mediastinum is poor. PET scanning has much better sensitivity and specificity than chest CT scanning for staging lung cancer in the mediastinum, and distant metastatic disease can be detected by PET scanning. With either test, abnormal findings must be confirmed by tissue biopsy to ensure accurate staging.

Figures in this Article

After a tissue diagnosis of lung cancer has been established or in patients in whom the clinical suspicion is high and surgery is the recommended next step, consideration must turn toward the determination of the extent of disease, or stage, because this will impact directly on management and prognosis. The most significant dividing line is between those patients who are candidates for surgical resection and those who are inoperable but will benefit from chemotherapy, radiation therapy, or both. Staging with regard to a patient’s potential for surgical resection is most applicable to non-small cell lung cancer (NSCLC); whereas, for small cell lung cancer (SCLC) a more simplified staging classification of limited and extensive disease is employed. Except in rare cases of surgically operable limited stage small cell cancer, the implication of staging on the management of SCLC is between chemotherapy and radiation for limited disease vs chemotherapy alone for extensive disease.1

The basis for staging NSCLC is the TNM system23 (see Table 1 for TNM descriptors and Figure 1 for stage grouping). From a practical standpoint, the involvement of disease in the mediastinum, reflected in the N designator in the system, most often determines appropriateness for surgical resection.

Patients with sage IA, IB, IIA, and IIB disease can benefit from surgical resection. Patients with stage IIIA, IIIB, and IV disease almost never meet the criteria for surgery. The current role of chemotherapy followed by surgery for selected patients with stage IIIA disease remains controversial.

Staging can be used to predict survival and to guide the patient toward the most appropriate treatment regimen or clinical trial. Even with clinical stage I, surgically resectable, potentially curable disease, the 5-year survival rate after surgery is only 50%. Approximately 60% of cancer recurrences are presumably from extrathoracic micrometastatic involvement at presentation, which is not currently detectable with existing diagnostic modalities. Patients with clinical stage II disease (T1N1M0 or T2N1M0) have a 5-year survival rate after surgery of 30%. At clinical stage IIIA, the 5-year survival rate is 17%, and at stage IIIB it is only 5%.3 These patients are generally treated with combined chemotherapy and radiotherapy. The 5-year survival rate for patients with stage IV disease is virtually nil, and this disease is treated either with chemotherapy and supportive care or with supportive care alone. Thus, one can see that it is critical to stage patients accurately as the treatment modalities and subsequent patient outcomes vary widely based on stage designation.

For this edition of the lung cancer guidelines, investigators from the Duke University Evidence-Based Practice Center and the authors of this guideline updated a systematic review of the diagnostic accuracy of noninvasive tests for staging in patients with NSCLC. The methods and results of the initial review have been published previously and a more complete description of the methodology can be found there.4 Briefly, the search strategy used computerized searches of the MEDLINE bibliographic database (January 1991 to May 2006), HealthStar, and the Cochrane Library. In addition, we searched the reference lists of included studies, selected textbooks, practice guidelines, systematic reviews, and metaanalyses in order to ensure that all relevant studies were identified. Only articles that had been published in English were considered.

Titles and abstracts, and the full text of all articles passing the title-and-abstract screen were evaluated independently by at least two of the authors for inclusion or exclusion based on the following five criteria: (1) publication in a peer-reviewed journal; (2) study size of 20 patients (except for studies involving CT scan evaluation of the mediastinum, for which 50 patients were required); (3) patient group not included in a subsequent update of the study; (4) histologic or cytologic confirmation of mediastinal nodes or extrathoracic sites in addition to the primary tumor; and (5) availability of the raw data needed to calculate independently the sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) of CT scanning, positron emission (PET) scanning, MRI, or endoscopic ultrasonography, or the raw data needed to calculate the NPV of the clinical evaluation.4

Recommendations were developed by the writing committee, graded by a standardized method (see the “Methodology for Lung Cancer Evidence Review and Guideline Development” chapter), and reviewed by all members of the lung cancer panel prior to approval by the Thoracic Oncology Network, Health and Science Policy Committee, and the Board of Regents of the American College of Chest Physicians.

Staging is a critical part of the evaluation of every patient with lung cancer. Defining malignant involvement of the mediastinal lymph nodes is particularly important, as the status of these nodes will in many cases determine whether there is surgically resectable disease. the clinical staging of lung cancer is usually directed by noninvasive imaging modalities. On the basis of such tests, clinicians will determine the likelihood of the presence or absence of tumor involvement in regional lymph nodes.

In general, patients with lung cancer can be separated into four groups with respect to intrathoracic radiographic characteristics (including both the primary tumor and the mediastinum), as shown in Figure 2 . Distinguishing these groups is particularly useful in defining the need for and selection of invasive staging tests. The first group (radiographic group A) involves patients with mediastinal infiltration that encircles the vessels and airways, so that discrete lymph nodes can no longer be discerned or measured. In these situations, the presence of mediastinal involvement (stage III disease) is generally accepted based on imaging studies alone, and the major issue is to obtain tissue by whatever approach is easiest in order to distinguish between SCLC and NSCLC. The second group (radiographic group B) involves patients with mediastinal node enlargement in whom the size of discrete nodes can be measured. In these patients, mediastinal nodal involvement is suspected but must be confirmed. The last two groups involve patients with normal mediastinal nodes. In radiographic group C, the presence of a central tumor or suspected N1 disease makes the chance of N2,3 nodal involvement relatively high (20 to 25%) despite normal-sized nodes, and further confirmation is needed.58 In the final group (ie, those patients with a peripheral clinical stage I tumor), the chance of mediastinal involvement is quite low, and generally further confirmation of this is not needed (radiographic group D).,68

A widely accepted definition of normal-sized mediastinal lymph nodes is a short-axis diameter of ≤ 1 cm on a transverse CT scan image. The term discrete nodal enlargement implies that discrete nodes are seen on the CT scan and are defined well enough to be able to measure their size (and are > 1 cm in size). Mediastinal infiltration is present when there is abnormal tissue in the mediastinum that does not have the appearance and shape of distinct lymph nodes, but instead has an irregular, amorphous shape. In this case, it is difficult to distinguish discrete nodes and impossible to come up with a measurement of the size of nodes. This occurs when multiple nodes are matted together to the point where the boundary between them is obscured, and can be assumed to involve extensive extranodal spread of the tumor. It may progress to the point where mediastinal vessels and other structures are partially or completely encircled. Finally, the distinction between a central tumor vs a peripheral tumor has also not been codified, but most authors consider any tumor in the outer two thirds of the hemithorax to be peripheral. Assessing the radiographic characteristics of the mediastinum will generally require that the clinician look at the images. This is because there is no standard format for how radiographic findings are reported (eg, the term lymphadenopathy is often used when there is a suspected malignancy, even though the mediastinal nodes are well below 1 cm in size).

The four radiographic groups are defined by anatomic characteristics seen on a CT scan (ie, size, location, and extent), and not by metabolic characteristics (ie, by PET scan) for many reasons. First, a CT scan is relatively inexpensive and essentially is always performed as a preliminary step in order to define the nature of a pulmonary abnormality and to arrive at a clinical diagnosis of suspected lung cancer. Second, the information gained from the clinical history, physical examination, and chest CT can define whether other tests such as a PET scan are indicated. Finally, the technical considerations and performance characteristics of invasive staging procedures are likely to be driven primarily by anatomic characteristics rather than by metabolic ones. In other words, the location and size of a lymph node are important in determining how feasible and reliable an invasive test is, and these issues are unaffected by whether the node in question is metabolically active on PET scanning or not. Further discussion of the best approach to confirming a diagnosis of mediastinal tumor involvement by tissue acquisition can be found in chapter 13 of this supplement on invasive staging.

The majority of lung cancers are initially detected on a plain chest radiograph. In some situations, the plain radiograph may be sufficient to detect spread of the tumor to the mediastinum. For example, the presence of bulky lymphadenopathy in the superior or contralateral mediastinal areas may be considered adequate evidence of metastatic disease, precluding a further imaging evaluation of the chest. This may be particularly true if the patient is too ill or is unwilling to undergo treatment of any kind. However, it is recommended that tissue confirmation be obtained if possible by the least invasive method available. It is widely accepted that the chest radiograph is in general an insensitive measure of mediastinal lymph node involvement with lung cancer; thus, further noninvasive and/or invasive assessment is usually necessary.

CT scanning of the chest is the most widely available and commonly used noninvasive modality for evaluation of the mediastinum in lung cancer. The vast majority of reports evaluating accuracy of CT scanning for mediastinal lymph node staging have employed the administration of IV contrast material. IV contrast is not absolutely necessary in performing chest CT scanning for this indication, but may be useful in helping to distinguish vascular structures from lymph nodes as well as in delineating mediastinal invasion by centrally located tumors. A CT scan of the chest should be performed in all cases of lung cancer unless the patient is so debilitated that no treatment is planned or they are unwilling to undergo further evaluation.

Various CT scan criteria have been used to define the malignant involvement of mediastinal lymph nodes. Notwithstanding the radiographic descriptions of mediastinal nodal involvement, the most widely used criterion is a short-axis lymph node diameter of ≥ 1 cm on a transverse CT scan. However, numerous other criteria have also been used including the following: (1) a long-axis diameter of ≥ 1 cm; (2) a short-axis diameter of ≥ 1.5 cm; (3) a short-axis diameter ≥ 1 cm plus evidence of central necrosis or disruption of the capsule; and (4) a short-axis diameter of ≥ 2 cm regardless of nodal morphology. The reported sensitivity and specificity for identifying malignant involvement will vary depending on which criteria are used in the assessment of individual nodal stations.910 The majority of studies evaluating CT scan accuracy have used a short-axis diameter of ≥ 1 cm as the threshold for abnormal nodes. In doing so, a conscious effort has been made to strike an appropriate balance between sensitivity and specificity in an understandable effort to minimize the number of false-positive evaluations without producing an unacceptable number of false-negative evaluations.

For the purposes of these guidelines, investigators from the Duke University Evidence-based Practice Center and the authors of this section of the supplement conducted a systematic review of the medical literature relating to the accuracy of CT scanning for noninvasive staging of the mediastinum in patients with lung cancer.4 Thirty-five studies published from 1991 through June 2006 evaluating the performance characteristics of CT scanning for this purpose were identified based on their fulfillment of the following criteria: (1) publication in a peer-reviewed journal; (2) a study size of > 50 patients; (3) patient group not included in a subsequent update of the study; (4) histologic or cytologic confirmation of mediastinal nodes or extrathoracic site as well as the primary tumor; and (5) availability of the raw data needed to calculate independently sensitivity, specificity, PPV, and NPV. These 43 studies6,1144,52,87,121122,178181 are outlined in Table 2 . The combined studies yielded 5,111 evaluable patients.,6,1144,52,87,121122,178181 The median prevalence of mediastinal metastasis was 28% (range, 18 to 56%). Almost all studies specified that CT scanning was performed following the administration of IV contrast material and that a positive test result was defined as the presence of one or more lymph nodes that measured > 1 cm on the short-axis diameter. Individual study estimates of sensitivity and specificity are shown in Figure 3 , which also displays the summary receiver operator characteristic (ROC) curve for mediastinal staging with CT scanning. ROC curves illustrate the tradeoff between sensitivity and specificity as the threshold that defines a positive test result varies from most to least stringent. The summary ROC method rests on the assumption that individual study estimates of sensitivity and specificity represent unique points on a common ROC curve. A summary ROC curve that lies closer to the upper left-hand corner of the diagram indicates better overall diagnostic accuracy. The pooled sensitivity and specificity of CT scanning for identifying mediastinal lymph node metastasis were 51% (95% confidence interval [CI], 47 to 54%) and 86% (95% CI, 84 to 88%), respectively. The corresponding positive and negative likelihood ratios were 3.4 and 0.6, respectively, confirming that CT scanning has a limited ability either to rule in or exclude mediastinal metastasis. The combined estimates should be interpreted with caution as the studies were statistically heterogeneous. Still, these findings mirror those of other analyses addressing the accuracy of CT scanning for staging the mediastinum in NSCLC. A large metaanalysis by Gould and colleagues,45reported the median sensitivity and specificity of CT scanning for identifying malignant mediastinal nodes as 61% and 79%, respectively, while an earlier metaanalysis by Dwamena and colleagues46 reported average sensitivity and specificity of 64% and 74%, respectively.

CT scanning is clearly an imperfect means of staging the mediastinum, but it remains the best overall anatomic study available for the thorax. A CT scan usually guides the choice of nodes for selective node biopsy by invasive techniques, and thus continues to be an important tool for diagnosing lung cancer. The choice of individual nodes for sampling as well as the choice of the most appropriate invasive technique (including transbronchial, transthoracic, or transesophageal needle aspiration, mediastinoscopy, or more extensive surgery) will typically be directed by the findings of the CT scan. However, the limitation of CT scan-based mediastinal lymph node evaluation is evident in the fact that 5 to 15% of patients with clinical T1N0 (clinical stage I) tumors will be found to have positive lymph node involvement by surgical lymph node sampling.47

Based on the currently available data relating to the performance characteristics of CT scanning for the evaluation of the mediastinum in patients with NSCLC, two important messages emerge. First, approximately 40% of all nodes that are deemed to be malignant by CT scan criteria are actually benign. Patient characteristics are a large factor, as specificity can be affected by clinical factors such as the presence of postobstructive pneumonitis.16 Second, approximately 20% of all nodes that are deemed to be benign by CT scan criteria are actually malignant. CT scanning can thus both overstage and understage the mediastinal nodes. In sum, there is no node size that can reliably determine tumor stage and operability. In cases in which the CT scan criteria for the identification of a metastatic node are met, the clinician must still prove beyond a reasonable doubt by biopsy or resection that the node is indeed malignant. Given the limitations of its imperfect sensitivity and specificity, it is usually inappropriate to rely solely on the CT scan to determine mediastinal lymph node status in patients with NSCLC. Nonetheless, CT scanning continues to play an important and necessary role in the evaluation of these patients. This conclusion is supported by the most recent American Thoracic Society/European Respiratory Society statement47on the pretreatment evaluation of NSCLC and British Thoracic Society guidelines48 on the selection of patients with lung cancer for surgery, both of which recommend CT scanning for the evaluation of mediastinal lymph nodes in all patients with suspected NSCLC. In the mediastinum, a CT scan can provide a road map that guides the location and modality to be used for subsequent biopsy procedures. In addition, patients with a very low pretest probability of metastasis (eg, those with small, peripheral T1 primary tumors) and no evidence of lymph node enlargement on a CT scan arguably might not require invasive staging prior to definitive thoracotomy. For example, when the clinical pretest probability is 10%, the posttest probability is approximately 6% when CT scan results are negative in the mediastinum.

1. For patients with either a known or suspected lung cancer who are eligible for treatment, a CT scan of the chest with contrast including the upper abdomen (liver and adrenal glands) should be performed. Grade of recommendation, 1B

2. In patients with enlarged discrete mediastinal lymph nodes on CT scans (> 1 cm on the short axis) and no evidence of metastatic disease, further evaluation of the mediastinum should be performed prior to definitive treatment of the primary tumor. Grade of recommendation, 1B

PET scanning is an imaging modality based on the biological activity of neoplastic cells. Lung cancer cells demonstrate increased cellular uptake of glucose and a higher rate of glycolysis when compared to normal cells.49The radiolabeled glucose analog 18F-fluoro-2-deoxy-D-glucose (FDG) undergoes the same cellular uptake as glucose and is phosphorylated by hexokinase, generating 18F-FDG-6-phosphate. The combination of increased uptake of 18F-FDG and a decreased rate of dephosphorylation by glucose-6-phosphatase in malignant cells results in an accumulation of 18F-FDG-6-phosphate in these cells.51 The concentrated isotope can then be identified using a PET camera. FDG-PET (subsequently referred to as PET) is thus a metabolic imaging technique that is based on the function of a tissue rather than its anatomy. Standardized quantitative criteria for an abnormal PET scan finding in the mediastinum are unfortunately lacking. A qualitative assessment is usually based on a comparison of uptake in the lesion or structure in question compared to the background activity of the lung or liver. A standard uptake value of < 2.5 is sometimes used as a threshold level for normalcy, but this measurement may vary with the new generation of scanners. Despite the lack of standardized criteria defining positive findings, PET scanning has proved useful in differentiating neoplastic from normal tissues. However, the technique is not infallible as nonneoplastic processes including granulomatous and other inflammatory diseases as well as infections may also demonstrate positive PET imaging findings. Further, size limitations are an issue, with the lower limit of spatial resolution of the current generation of PET scanners being approximately 7 to 10 mm. However, smaller lesions may be detected, depending on the intensity of uptake of the isotope in abnormal cells.30,52Additionally, certain well-differentiated low-grade malignancies, particularly bronchioloalveolar cell carcinoma and typical carcinoid tumors, are known to have higher false-negative finding rates.5357

A burgeoning number of studies in the last several years have reported on the utility of PET scanning in the assessment of the mediastinum in patients with lung cancer. The increasing availability of the technology now allows PET scanning to be used widely as a diagnostic tool. It should be noted that PET scanning is primarily a metabolic examination and has limited anatomic resolution. It is usually possible by PET scanning to identify lymph node stations, but not individual lymph nodes. CT scanning provides much more anatomic detail but lacks the functional information provided by PET scanning. Newer generation integrated PET-CT imagers may combine the advantages of both studies, but there are as yet few studies addressing the accuracy of this modality.58

As was done for CT scanning, investigators from the Duke University Evidence-based Practice Center performed a systematic review4 of the medical literature relating to the accuracy of PET scanning for noninvasive staging of the mediastinum in patients with lung cancer. Studies evaluating the performance characteristics of PET scanning for this purpose were identified based on their fulfillment of the following criteria: (1) publication in a peer-reviewed journal; (2) study size of > 20 patients; (3) patient group not included in a subsequent update of the study; (4) histologic or cytologic confirmation of mediastinal nodes or extrathoracic site as well as the primary tumor; and (5) availability of the raw data needed to calculate independently sensitivity, specificity, PPV, and NPV. All studies were interpreted in conjunction with patients’ CT scan findings so that the PET scan findings were correlated with the anatomic location of the lesion seen on the CT scan. In all studies, 18F-FDG was the radiopharmaceutical used for imaging. Forty-four studies ,6,8,1112,20,22,2425,2728,30,33,35,37,39,42,44,52,5978,87,121122,178,182183 published between 1994 and June 2006 were identified, yielding 2,865 evaluable patients. These studies are displayed in Table 3 . The median prevalence of mediastinal metastasis was 29% (range, 5 to 64%). Figure 4 shows individual study estimates of sensitivity and specificity and the summary ROC curve for the PET scans. Pooled estimates of sensitivity and specificity for identifying mediastinal metastasis were 74% (95% CI, 69 to 79%) and 85% (95% CI, 82 to 88%), respectively. Corresponding positive and negative likelihood ratios for mediastinal staging with PET scanning were 4.9 and 0.3, respectively. These findings demonstrate that PET scanning is more accurate than CT scanning for staging of the mediastinum in patients with lung cancer, though it is far from perfect.

PET scanning may provide an additional benefit in that it is a whole-body study. The usual extrathoracic staging of lung cancer will typically include a combination of bone scintigraphy, brain imaging by CT scanning or MRI and abdominal CT scanning or the inclusion of the upper abdomen in a chest CT scan. PET scanning is able to provide information about the primary site in the chest as well as intrathoracic and extrathoracic metastases with a single study. The exception to this is the definition of metastases in the brain, as the brain will normally avidly take up 18F-FDG. Several studies30,42,79 have reported on the ability of PET scanning to identify extrathoracic metastases in patients whose tumors had been deemed resectable by conventional imaging. The rate of detection of unanticipated M1 disease by PET scanning has been reported as 1 to 8% in patients with clinical stage I disease and 7 to 18% in patients with clinical stage II disease.42,79The identification of unanticipated distant metastases by PET scanning in such patients should result in the avoidance of unwarranted thoracotomies, but all positive findings in surgical candidates should be confirmed by biopsy unless there is overwhelming evidence of distant metastasis.80

To summarize, PET scanning has both higher sensitivity and higher specificity than CT scanning for the evaluation of mediastinal lymph nodes, and can provide important information regarding the presence of metastatic disease outside the thorax. In the mediastinum, PET scanning is more accurate than CT scanning in identifying abnormal nodes that can be sampled by directed biopsy. Accordingly, PET scanning has assumed an increasingly important role in the evaluation of patients with lung cancer. However, broader experience with PET scanning has not yet allowed a precise definition of its role in the staging evaluation of lung cancer. PET scanning is not infallible. False-positive PET scan findings may result in missed opportunities for a cure by surgical resection. Conversely, false-negative PET scan findings may lead to fruitless thoracotomies in patients with unresectable disease. The potential consequences of both false-positive and false- negative PET scan findings in an environment in which PET scanning is increasingly relied on for staging must be considered when PET scanning is included in the evaluation of NSCLC.

Some studies45,8183 have pointed out that the accuracy of PET imaging in the mediastinum is dependent on the size of the nodes identified by CT scanning. PET scanning is more sensitive (but less specific) when CT scanning identifies enlarged nodes.45,81 In a metaanalysis evaluating the conditional test performance of PET and CT scanning, Gould and colleagues45 reported median sensitivity and specificity of PET scans of 100% and 78%, respectively, in patients with enlarged lymph nodes. PET scanning is thus very accurate in identifying malignant nodal involvement when nodes are enlarged. However, PET scanning will falsely identify malignancy in approximately one-fourth of patients with nodes that are enlarged for other reasons, usually inflammation, or infection. Positive PET findings in this situation should be confirmed by directed biopsy. Failure to do so could result in patients with surgically resectable disease being denied curative surgery. An argument could also be made that a patient in whom the clinical assessment of pretest probability of malignant node involvement is high should proceed directly to biopsy without PET, as a negative PET result would not negate a strong clinical suspicion for tumor. In this situation, negative PET findings would be unlikely to change the clinical suspicion for malignancy enough to defer histologic confirmation. As a counter-argument, PET scanning might still impact the decision process if unexpected extra-thoracic sites of abnormal activity are found, and patients with clinical stage III disease are at highest risk for occult distant metastasis. Identification of such foci might affect the choice of biopsy site and have a significant impact on the clinical stage and the decision of whether a patient should undergo surgical resection. Whether this is adequate reason to pursue PET scanning in patients with enlarged mediastinal nodes by CT scanning in whom the clinical suspicion for malignant involvement is high is unanswered.

Conversely, PET scanning is less sensitive (but more specific) in patients with normal-sized mediastinal nodes seen by CT scanning. Based on the data presented in Table 2, CT scanning of the mediastinum is falsely negative in approximately 20% of patients with normal-sized nodes and malignant nodal involvement. In the metaanalysis reported by Gould and colleagues,45 the median sensitivity and specificity of PET scanning in this group of patients were 82% and 93%, respectively. These data indicate that nearly 20% of patients with normal-sized nodes but with malignant involvement had falsely negative PET scan findings. Corresponding positive and negative likelihood ratios were approximately 12.0 and 0.2, respectively. In this study, when both CT and PET scan results were negative and the pretest probability of mediastinal lymph node metastasis was estimated at 35% (which corresponds to the median prevalence of mediastinal metastasis in studies of PET scanning), the posttest probability of mediastinal metastasis was approximately 9% (95% CI, 4 to 14%). This addresses the controversial question of whether a negative PET scan finding in patients with normal-sized lymph nodes by CT scanning can obviate the need to perform further invasive mediastinal evaluation prior to thoracotomy. In this situation, we believe that the appropriate invasive staging procedure would be mediastinoscopy, as there are no enlarged nodes to directly biopsy by other techniques. While PET scanning samples all mediastinal nodal groups, it is clearly less sensitive for nodes with a diameter of < 7 to 10 mm. While mediastinoscopy cannot sample all mediastinal nodal groups, it can detect microscopic disease even in small nodes. Ultimately, the decision as to whether a negative PET scan finding can be used to obviate mediastinoscopy will require clinical judgment that incorporates multiple factors, including the clinical pretest probability of mediastinal metastasis, patient preferences, and local availability and expertise in both mediastinoscopy and PET imaging (see the “Invasive Mediastinal Staging of Lung Cancer” chapter for further recommendations).

The utility of PET scanning in patients with stage 1A disease is less clear as the prevalence of mediastinal and distant metastatic disease is low and the evidence for utilizing PET scanning is poor. Further study in this specific patient population is warranted prior to making a recommendation that has a higher level of evidence.

In summary, PET scanning is the most accurate noninvasive imaging modality available to evaluate the mediastinum in patients with lung cancer. Abnormal findings on PET scans may be important in identifying mediastinal nodes for directed biopsy. PET scanning is also a whole-body study and offers additional information relating to extrathoracic sites of possible disease involvement (see “The Search for Metastatic Disease” section). However, wider experience with PET scanning has increased the awareness of the potential for and consequences of both false-positive and false-negative findings.

3. PET scanning to evaluate for mediastinal and extrathoracic staging should be considered in patients with clinical 1A lung cancer being treated with curative intent. Grade of recommendation, 2C

4. Patients with clinical 1B-IIIB lung cancer being treated with curative intent, should undergo PET scanning (where available) for mediastinal and extrathoracic staging. Grade of recommendation, IB

5. In patients with an abnormal result on FDG-PET scans, further evaluation of the mediastinum with sampling of the abnormal lymph node should be performed prior to surgical resection of the primary tumor. Grade of recommendation, 1B

An important shortcoming of dedicated PET imaging is its limited spatial resolution, which results in poor definition of anatomic structures. As a result, it may be difficult for PET scanning to distinguish between mediastinal and hilar lymph nodes, or to differentiate between a central primary tumor and a lymph node metastasis, even when the results of PET and CT scans are visually correlated. This limitation has been addressed by the development of “dual-modality” or “integrated” PET/CT scanning systems, in which a CT scanner and a PET scanner are combined in a single gantry. Some studies2425,58,8485 have begun to examine the accuracy of integrated PET/CT scanners for lung cancer staging. The total number of patients evaluated by this hybrid technique is still relatively small. Estimates of accuracy for identifying mediastinal metastasis are limited, though early studies have indicated2425,85 that the sensitivity and specificity are at least as good as those with PET scanning alone.

Like CT scanning, MRI is an anatomic study. Data relating to the accuracy of the evaluation of the mediastinum with MRI in patients with NSCLC are limited, but available reports13,86 suggest that the accuracy of MRI is as good as CT scanning. Two reports8687 also have suggested that the use of contrast enhancement may improve the accuracy of MRI in this situation. MRI may be superior to CT scanning for defining lung cancer spread in the thorax in specific situations. Because MRI can detect differences in intensity between tumor and normal tissues, including bone, soft tissues, fat, and vascular structures, it may be more accurate than CT scanning in delineating direct tumor invasion of the mediastinum, chest wall, diaphragm, or vertebral bodies.13,8891 This may be particularly useful in evaluating superior sulcus tumors or tumors abutting the mediastinum, structures of the chest wall, and diaphragm. However, most centers continue to rely on CT scanning as the noninvasive anatomic study of choice for evaluating potential mediastinal spread of lung cancer.

6. For patients with either a known or suspected lung cancer who are eligible for treatment, an MRI of the chest should not routinely be performed for staging the mediastinum. MRI may be useful in patients with NSCLC where there is concern for involvement of the superior sulcus or brachial plexus involvement. Grade of recommendation, 1B

The purpose of extrathoracic scanning in patients with NSCLC is usually to detect metastatic disease, especially at common metastatic sites such as the adrenal glands, liver, brain, and skeletal system, thereby sparing the patient fruitless radical treatment.92However, scans can only detect macroscopic metastatic deposits that have reached a size within the resolution capability of the imaging modality in question, and this can be considered a major shortcoming of all conventional tests currently used to detect distant metastases in patients with NSCLC. In more recent years, increasing attention has focused on the use of immunocytochemical techniques using monoclonal antibodies to detect occult micrometastases, which are sometimes associated with a worse prognosis, in the bone marrow of NSCLC patients.9398 Such techniques may add a new dimension to metastatic staging in the near future.

In the meantime, the preferred scans for staging patients with NSCLC in 2007 are CT scanning of the chest, CT scanning or MRI with contrast of the brain, and 99Tc nuclear imaging of the skeletal system. The use of whole-body PET scans for extrathoracic staging is evolving, and PET scanning may ultimately play a significant role in the assessment of distant disease. The very limited extant data regarding whole-body single photon emission CT scanning for metastatic disease suggest that its performance is slightly inferior to that of PET scanning.72,79

It is clear that the use of extrathoracic scans must always be subordinate to a thoughtful overall clinical strategy for each individual patient. For example, a whole-body PET scan has little role in the diagnosis of a patient with clinically obvious, accessible advanced disease, such as skin metastases or massive hepatic replacement by metastatic tumor seen on CT scans.5354,99 In other circumstances, the need for tissue confirmation of metastatic disease can supercede the need for additional sophisticated scanning. For instance, in certain patients an adrenal biopsy, rather than a PET scan, may be required to clarify the nature of a unilateral adrenal mass seen on a CT scan.

It is well established that abnormal symptoms, physical examination findings, and routine blood tests in the initial clinical evaluation of patients with NSCLC are associated with a significant yield (approximately 50%) of abnormal scan findings.92 Moreover, a rough semiquantitative relationship has been demonstrated in some studies92,100 between the number of abnormal “clinical factors” and the frequency of abnormal scan findings. In the absence of all clinical factors, the scan yield is much lower, giving rise to the recommendation that scans be omitted in this setting,31,48,100104 though controversy persists on this point.105Other important variables focus on the primary lesion, since more scan abnormalities are associated with advanced thoracic lesions (T and N factors).106107 This is particularly true for patients with N2 disease, in whom asymptomatic metastases have been documented at a higher rate than would have been expected.106107 There has been some controversy with regard to cell type and the incidence of asymptomatic metastases. Several studies108109 have documented a higher incidence of brain metastases with adenocarcinomas as opposed to squamous cell cancers, but a large series104 of patients with stage I and II lung cancer found no difference.

Several important caveats pertain to scanning for distant metastases in general. First is the issue of false-positive scan findings. Clinical entities that frequently give rise to false-positive scan findings include adrenal adenomas (present in 2 to 9% of the general population), hepatic cysts, degenerative joint disease, old fractures, and a variety of nonmetastatic space-taking brain lesions. When clinically indicated, additional imaging studies and/or biopsies are performed to establish the diagnosis, but complications and costs resulting from such subsequent investigations have received insufficient attention.110111 A second problem is that of false-negative scan findings (ie, metastases that are present but not picked up by current scanning techniques). This was demonstrated convincingly by Pagani,112 who found metastatic NSCLC in 12% of radiologically normal adrenal glands by percutaneous biopsy; a more recent autopsy series113 suggested that the sensitivity of CT scanning for adrenal metastases may be as low as 20%. A third difficulty is that most studies fail to carefully specify exactly which elements comprise the prescan clinical evaluation, or invoke differing clinical indicators to mandate scanning. Organ-specific findings such as headache and non-organ-specific complaints such as weight loss are both important.100,114 The current preferred “expanded” clinical evaluation includes organ-specific and constitutional signs and symptoms, along with simple laboratory test results, as shown in Table 4 .,92 Furthermore, Guyatt et al115have shown that careful delineation and quantification of historical features using a 5-point scale of severity can importantly affect the subsequent scan yield and ultimately the incidence of metastases after lung cancer surgery. A fourth issue is an ascertainment problem, since abnormal scan findings in many studies were not followed up with definitive biopsy proof of metastatic disease. This may relate to anatomic factors, overall debility, or refusal of the patient, or a variety of other cogent clinical concerns. Fifth, it must be noted that even biopsy proof of metastatic disease does not dictate a certain clinical management pathway. Carefully selected patients with localized lung cancers in the thorax, accessible, solitary metastases to the brain or adrenal gland, and other favorable clinical features may obtain long-term survival with an aggressive treatment approach, including surgical extirpation of both the primary and metastatic site.116117 Finally, the lack of prospective randomized trials and outcome studies in the area of extrathoracic staging is striking. Two retrospective studies showed that scanning asymptomatic patients with early NSCLC did not help to predict recurrences postoperatively or to improve survival.118119 The only prospective randomized trial120 showed no statistical difference in recurrence rates or survival in a group of patients who were randomized to undergo bone scintigraphy and CT scans of the head, liver, and adrenal glands, compared with the group assigned to undergo CT scans of the chest and mediastinoscopy, followed by thoracotomy when appropriate.

Utility of PET Scanning for Detecting Metastatic Disease

Since 1993, numerous studies have assessed the clinical utility of PET scans to assist in the search for metastatic disease in patients with NSCLC. In general, these tend to be relatively small, prospective, single-institution assessments in which whole-body PET scanning suggests the presence of unsuspected distant disease in 10 to 20% of cases.20,27,121122 The yield of unsuspected metastases depends on a number of factors, including whether PET scanning is gauged as an initial metastatic evaluation, or only after some metastases have already been detected via conventional scans.42,123 The yield is higher in patients with clinical stage III disease,79 and a relationship between thoracic nodal stage and PET scanning yield has been suggested.124When the area of interest is a single site (eg, adrenal glands or skeletal system), the performance characteristics (ie, sensitivity, specificity, PPV, NPV, and accuracy) of PET scanning are very favorable, often surpassing the performance of conventional imaging with CT scans or radionuclide bone imaging.126 Furthermore, whole-body PET scanning enables the imaging of areas not covered in the traditional scanning algorithm, allowing the detection of occasional metastatic foci in, for example, skin, pelvis, skeletal muscle, soft tissue, kidney, and pancreas.27 In most of these studies, abnormal PET scan findings are followed up with biopsy, serial conventional radiographs, and/or careful clinical assessment to confirm the veracity of the PET scan findings.

Nevertheless, several concerns pertain specifically to the emerging literature regarding PET scans as a test for distant disease. First, the exact criterion for a positive PET scan finding is usually based on an entirely subjective or semiquantitative comparison with background activity. Attempts to derive a reliable criterion based on standardized or differential uptake ratios have been generally unsuccessful to date. Second, several significant problems attend the use of PET scanning as an imaging modality for brain metastases. Not only does high baseline brain uptake pose a problem in detecting focal accumulations,28 but many PET scanners include only the area from the base of the skull to the mid-thighs, thereby excluding much of the brain parenchyma from the images. Obtaining satisfactory brain PET scan images can require special equipment modifications and prolonged image-acquisition time.20,127 Furthermore, the small size of most brain metastases may be problematic in terms of the limited resolution of conventional PET scans. Third, while there is some evidence that PET scanning can avert unnecessary thoracotomies,80 improve clinical staging,20,121122,128 influence patient management decisions,128 and alter radiotherapy planning,79 there has been scant evidence to date linking PET scanning to an improvement in important patient outcomes such as recurrences of metastatic disease or mortality, and cost-effectiveness assessments are just beginning to emerge.123,129130 Fourth, a substantial ascertainment problem exists for negative PET scan findings, in that metastatic disease missed by PET scanning is generally unverifiable; thus, the false-negative rate is not truly knowable in most studies. But in one study,131 19% of patients who underwent a curative resection experienced a systemic relapse within a mean interval of 14 months despite a negative finding on a preoperative whole-body PET scan, suggesting that the false-negative problem may be significant. Finally, some of the larger, more recent multiinstitutional studies42 have shown substantially lower performance characteristics for PET scanning than those in the initial studies, with a PPV as low as 36% for metastatic disease.

To some extent, the very recent tempering of enthusiasm for PET scanning for distant disease likely reflects the usual trajectory of a new test, as greater experience accumulates in thousands of patients under a wide variety of clinical circumstances and interpretive expertise. In this sense, the experience with PET scanning echoes the experience with CT scanning of the mediastinum in patients with NSCLC, in which initial reports of sensitivity and specificity were in excess of 90%, before settling into the accepted values of 60 to 70% decades later. On the other hand, more recently introduced integrated PET/CT scanners offer the hope of combining metabolic imaging with precise anatomic resolution to further refine the search for metastatic disease.58,84,132 In one highly publicized study,58 integrated PET-CT scanning increased diagnostic certainty as to the precise location of metastasis in two of eight patients in whom conventional PET scanning detected unsuspected extrathoracic focal accumulations.

Thus, it is premature to definitively assess the role of whole-body PET scanning in the search for metastatic disease barely 10 years after its introduction into clinical practice. As of this writing, it appears that whole-body PET scanning is best suited to help resolve cases in which prior imaging of a possible metastatic deposit is equivocal, and to detect unsuspected distant metastasis in either the preoperative setting or in those patients who are at high risk for metastatic deposits even when they are clinically asymptomatic (clinical stage IIIA).131

Detection of Abdominal Metastases

Some PET scan studies can also be considered in the context of the scanning of individual organ systems in patients with NSCLC. Thirteen studies105107,109,133141 evaluated the utility of clinical evaluation in detecting abdominal metastases in 1,291 patients using CT scanning as the reference standard (Table 5 ). Most of the studies limited study enrollment to patients with a negative clinical evaluation. In these nine studies,,107,109,133137,139140 the median prevalence of abdominal metastasis was 3% (range, 0 to 18%), and the median predictive value of a negative clinical evaluation was 97% (range, 82 to 100%). Four studies105106,138,141 enrolled patients with both positive and negative clinical evaluation findings. In these studies, the prevalence of abdominal metastasis ranged between 6% and 40%. Both sensitivity (range, 40 to 100%) and specificity (range, 27 to 65%) varied widely across studies. The use of CT scanning as an imperfect reference standard suggests that these estimates should be interpreted with caution.

It is relatively common to encounter adrenal masses on a routine CT scan, but many of these lesions are unrelated to the malignant process. A unilateral adrenal mass in a patient with NSCLC is more likely to be a metastasis than a benign lesion according to some studies,92,142but not others.143144 In the presence of clinical T1N0 NSCLC, adenomas predominate,135136 whereas adrenal metastases are frequently associated with large intrathoracic tumors or other extrathoracic metastases.92,145 Many studies140 have suggested that the size of a unilateral adrenal abnormality seen on a CT scan is an important predictor of metastatic spread, but this has not been a universal finding.

PET scans have performed exceptionally well in several studies specifically addressing the problem of adrenal metastases in NSCLC, with accuracy as high as 100% in two studies.28,146 However, small lesions (< 15 mm) were underrepresented in these series, and other studies have noted rare false-positive findings in this site.30,125,131

Four possible approaches to distinguishing between malignant and benign adrenal masses have been proposed, as follows: evaluation by specific CT scanning or MRI criteria; evaluation with additional or serial imaging; evaluation by percutaneous biopsy; and evaluation by adrenalectomy. Well-defined, low-attenuation (fatty) lesions with a smooth rim on unenhanced CT scan are more likely to be benign adenomas,147149 but the CT scan appearance of many lesions is insufficiently distinctive.147 Follow-up scanning with repeat CT, serial ultrasounds, MRI (especially with chemical shift and dynamic gadolinium-enhanced techniques150), 131-6-betaiodomethylnorcholesterol scanning,151 or PET scanning can often help with the critical distinction between metastatic disease and adenoma. Percutaneous adrenal biopsy is a relatively safe and effective means of achieving a definitive diagnosis in doubtful cases, and is especially important when the histology of the adrenal mass will dictate subsequent management.133134 However, this procedure may be nondiagnostic or unfeasible due to anatomic constraints. When insufficient material results from a biopsy, repeat aspiration or even adrenalectomy should be considered.140,147

Most liver lesions are benign cysts or hemangiomas, but a contrast CT scan (or ultrasound) is often required to establish a likely diagnosis.47 Percutaneous biopsy can be performed when diagnostic certainty is required. One metaanalysis110 that specifically reviewed hepatic studies derived a pooled yield of 3% for liver metastases in asymptomatic patients with NSCLC. PET scanning can detect liver metastases with an accuracy of 92 to 100% and only rare false-positive findings, though data in patients with NSCLC are very limited at present.20,28

Detection of Brain Metastases

In most studies, the yield of CT scanning/MRI of the brain in NSCLC patients with negative clinical examination findings is 0 to 10%,152158 possibly rendering the test cost-ineffective.154 Eighteen studies31,100,105,107109,137,152153,155156,158164 evaluated the ability of clinical evaluation to detect brain metastases in comparison to CT in 1,830 patients (Table 6 ). Nine studies,31,107,137,152153,155,158159,164 limited enrollment to patients with a negative clinical evaluation. In these studies, the median prevalence of brain metastasis was 3% (range, 0 to 21%), and the median predictive value of a negative clinical evaluation finding was 97% (range, 79 to 100%). Nine other studies100,105,108109,156,160163 enrolled patients with both positive and negative clinical evaluation findings. In these studies, the median prevalence of brain metastasis was higher (14%; range, 6 to 32%). The pooled sensitivity and specificity were 76% (95% CI, 61 to 87%) and 82% (95% CI, 69 to 91%), respectively.

An association among brain metastases, N2 disease in the chest, and adenocarcinoma histology has been described.108,157158 The rate of false-negative findings on CT scans wherein patients return with brain metastases within 12 months of the original scan is reported to be 3%.158 False-positive scan results can be a problem in up to 11% of patients due to brain abscesses, gliomas, and other lesions165; therefore, biopsy may be essential in patients in whom management is critically dependent on the histology of the brain lesion.

MRI is more sensitive than CT scanning of the brain and picks up more lesions and smaller lesions,166 but in some studies164 this has not translated into a clinically meaningful difference in terms of survival. While studies show that MRI can identify additional lesions in patients with metastases, there are no studies that show that MRI is able to identify more patients with metastases from lung cancer compared to CT scanning. Therefore, CT scanning is an acceptable modality for evaluating patients for metastatic disease. If the primary lesion is more advanced than T1N0M0, MRI with contrast can identify asymptomatic, verifiable metastases to the brain in 22% of patients with NSCLC and surgically resectable thoracic disease.167However, the use of routine MRI in staging NSCLC patients with negative clinical evaluation findings has not been adequately studied to date; a role in patients with large cell carcinoma or stage III adenocarcinoma has been suggested.168

Many of the shortcomings of PET scans in imaging the brain have been alluded to. In addition, performance has been suboptimal, with sensitivity as low as 60%,28 and occasional false-negative imaging findings of even sizable brain metastases.169 One study30 has suggested that PET scanning with 11C-labeled choline may be far superior to the usual 18F-FDG PET scanning for imaging brain metastases. In general, PET scanning is not considered to be reliable for detecting brain metastases.

Detection of Bone Metastases

The problem of false-positive scan abnormalities in radionuclide bone scintigraphy is particularly nettlesome, owing to the frequency of degenerative and traumatic skeletal damage and the difficulty in obtaining a definitive diagnosis via follow-up imaging or biopsy. False-positive bone imaging findings also occur with MRI, which may be no more accurate than nuclear bone imaging.167 Eight studies examined the ability of the clinical evaluation to detect bone metastases in 723 patients using bone scanning as the reference standard (Table 7 ).,101103,105,109,137,170171 Two studies102,137 limited enrollment to patients with negative clinical evaluation findings. In one study102 that included patients with both SCLC and NSCLC, the prevalence and NPV were 16% and 84%, respectively. In a subsequent study137 of patients with NSCLC, the prevalence and NPV were 30% and 70%, respectively. Six studies101,103,105,109,170171 enrolled patients with both positive and negative clinical evaluation findings. In these studies, the median prevalence of bone metastasis was 16% (range, 8 to 27%), and the pooled sensitivity and specificity were 87% and 67%, respectively.

Using radionuclide bone scanning as the reference standard, the pooled negative predicted value of the clinical assessment was 90% (95% CI, 86 to 93%). The relatively high frequency of unsuspected positive scan findings has led some investigators170 to recommend routine bone scanning in all preoperative patients. This concept is supported by the results of a study172 in which 27% of asymptomatic patients were found to have skeletal metastases. False-negative findings on a bone scan can also be a problem, and in one series171 skeletal metastases developed within 1 year in 6% of patients who had an initially negative bone scan result. PET scanning appears to have excellent performance characteristics in assessing bone metastases, with specificity, sensitivity, NPV, PPV, and accuracy all exceeding 90%,28,126 though false-positive and false-negative findings are occasionally seen.28,42,131 The accuracy of PET scanning surpassed that of radionuclide bone scanning in two direct comparative studies.172173

Pleural/Lung Metastases

The limited data suggest that PET scanning can be useful in identifying lung metastases28,174and malignant pleural effusions175176 in NSCLC patients, though much of the data pertains to nonpulmonary malignancies. False-positive and false-negative findings have occasionally been noted.30,175,177178

7. For patients with either a known or suspected lung cancer, a thorough clinical evaluation similar to that listed in Table 4 should be performed. Grade of recommendation, 1B

8. Patients with abnormal clinical evaluations should undergo imaging for extrathoracic metastases. Site-specific symptoms warrant a directed evaluation of that site with the most appropriate study (eg, head CT scanning/MRI plus either whole-body PET scanning or bone scanning plus abdominal CT scanning). Grade of recommendation, 1B

9. Routine imaging for extrathoracic metastases (eg, head CT scanning/MRI plus either whole-body PET scanning or bone scanning plus abdominal CT scanning) should be performed in patients with clinical stage IIIA and IIIB disease (even if they have negative clinical evaluation findings). Grade of recommendation, 2C

10. Patients with imaging study findings that are consistent with distant metastases should not be excluded from potentially curative treatment without tissue confirmation or overwhelming clinical and radiographic evidence of metastases. Grade of recommendation, 1B

CT scanning of the chest is useful in providing anatomic detail that better identifies the location of the tumor, its proximity to local structures, and whether or not lymph nodes in the mediastinum are enlarged. Unfortunately, the accuracy of chest CT scanning in differentiating benign from malignant lymph nodes in the mediastinum is unacceptably low. Whole-body PET scanning provides functional information on tissue activity, and has much better sensitivity and specificity than chest CT scanning for staging lung cancer in the mediastinum. In addition, distant metastatic disease can be detected by PET scanning. Still, positive findings on PET scans can occur as a result of nonmalignant etiologies (eg, infections), so tissue sampling to confirm suspected metastasis is usually required.

The clinical evaluation tool, that is, a thorough history and physical examination, remains the best predictor of distant metastatic disease. If the clinical evaluation finding is negative, then imaging studies such as CT scans of the head, bone scans, or abdominal CT scans are unnecessary and the search for metastatic disease is complete. If the signs, symptoms, or findings from the physical examination suggest malignancy, then sequential imaging, starting with the most appropriate study based on the clues obtained by the clinical evaluation, should be performed.

Abnormalities detected by any of the aforementioned imaging studies are not always cancer. Unless overwhelming evidence of metastatic disease is present on an imaging study, and where it will make a difference in treatment, all abnormal scan findings require tissue confirmation of malignancy so that patients are not denied the opportunity to have potentially curative treatment.

1. For patients with either a known or suspected lung cancer who are eligible for treatment, a CT scan of the chest with contrast including the upper abdomen (liver and adrenal glands) should be performed. Grade of recommendation, 1B

2. In patients with enlarged discrete mediastinal lymph nodes seen on CT scans (ie, > 1 cm on the short axis) and no evidence of metastatic disease, further evaluation of the mediastinum should be performed prior to definitive treatment of the primary tumor. Grade of recommendation, 1B

3. PET scanning to evaluate for mediastinal and extrathoracic staging should be considered in patients with clinical 1A lung cancer being treated with curative intent. Grade of recommendation, 2C

4. Patients with clinical 1B-IIIB lung cancer being treated with curative intent, should undergo PET scanning (where available) for mediastinal and extrathoracic staging. Grade of recommendation, IB

5. In patients with an abnormal result on FDG-PET scans, further evaluation of the mediastinum with sampling of the abnormal lymph node should be performed prior to surgical resection of the primary tumor. Grade of recommendation, 1B

6. For patients with either a known or suspected lung cancer who are eligible for treatment, an MRI of the chest should not be routinely performed for staging the mediastinum. MRI may be useful in patients with NSCLC in whom there is concern for involvement of the superior sulcus or brachial plexus. Grade of recommendation, 1B

7. For patients with either a known or suspected lung cancer, a thorough clinical evaluation similar to that listed in Table 4 should be performed. Grade of recommendation, 1B

8. Patients with abnormal clinical evaluation findings should undergo imaging for extrathoracic metastases. Site-specific symptoms warrant a directed evaluation of that site with the most appropriate study (eg, head CT scanning/MRI plus either whole-body PET scanning or bone scanning plus abdominal CT scanning). Grade of recommendation,1B

9. Routine imaging for extrathoracic metastases (eg, head CT scanning/MRI plus either whole-body PET scanning or bone scanning plus abdominal CT scanning) should be performed in patients with clinical stage IIIA and IIIB disease (even if they have a negative clinical evaluation finding). Grade of recommendation, 2C

10. Patients with imaging study findings that are consistent with distant metastases should not be excluded from potentially curative treatment without tissue confirmation or overwhelming clinical and radiographic evidence of metastases. Grade of recommendation, 1B

Abbreviations: CI = confidence interval; FDG = fluoro-2-deoxy-D-glucose; NPV = negative predictive value; NSCLC = non-small cell lung cancer; PET = positron emission tomography; PPV = positive predictive value; ROC = receiver operating characteristic; SCLC = small cell lung cancer

Table Graphic Jump Location
Table 1. TNM Descriptors