In the past few decades, the use of computers has revolutionized imaging, with the introduction of technologies such as computed tomography (CT), magnetic resonance imaging (MRI), ultrasonography, positron emission tomography (PET), and, more recently, PET-CT, which integrates anatomic (morphologic) and physiologic aspects of imaging. With the ever-greater subspecialization of the different areas of practice within medical oncology—surgical oncology, radiation oncology, and diagnostic radiology—and the expanding use of picture archiving systems, radiologist and clinician may encounter each other only rarely, if at all. Optimal patient outcomes, however, require careful planning of imaging for diagnosis, staging, and follow-up, best achieved through direct communication between the clinician and the radiologist. This chapter presents a general overview of lung and pleural tumor imaging, with an emphasis on the strengths and weaknesses of specific techniques in evaluating different tumor types, to help in selection of the ideal imaging modality for each patient.
PRIMARY MALIGNANT LUNG TUMORS
Screening
Despite new diagnostic techniques, the overall 5-year survival rate for patients with lung cancer, the leading cause of cancer death, remains approximately 15%, and most patients still present with advanced disease. This high death rate is presumed to reflect a combination of difficulty in detecting early-stage disease and lack of significant curative treatment. Abrogating cigarette smoking would be highly effective in reducing the prevalence of lung cancer but would not abolish it altogether, because effecting lifestyle change in an entire population is very difficult; moreover, previous smokers would still be at risk for lung cancer. Detection of the disease at the stage at which cure or control is possible is the theoretical rationale for screening for lung cancer.
Because tumors of the lungs are encased by the rib cage, early diagnosis by physical examination is not possible. Chest radiographs are ideal for demonstrating pulmonary abnormalities that differ significantly from the surrounding structures in density. The lungs contain air, the density of which differs significantly from the soft tissue density of tumor. Early screening studies for lung cancer, therefore, used chest radiography, which fulfills the criteria for a suitable screening test by being simple to perform, inexpensive, painless, and relatively safe, with relatively limited radiation exposure. Nonrandomized, uncontrolled screening studies in the 1950s gave way to nonrandomized, controlled trials, which showed that persons in the screened group were more likely to have lung cancer detected in the early stages, were more likely to have resectable disease, and enjoyed better survival rates. No clear reduction in lung cancer–associated mortality, however, was documented.
Although survival (number of persons alive after diagnosis of the disease relative to the total number of persons diagnosed with the disease) is commonly reported in screening trials, this statistic can be misleading because it is subject to lead time, study duration, and overdiagnosis biases. An impact on mortality rather than survival is therefore sought, to validate potential screening methods. Accordingly, in the 1970s, four major randomized, controlled trials looked at approximately 37,000 male smokers and found that chest radiograph screening yielded no change in mortality. In the screened cohort, patients demonstrated higher 5-year survival rates but no reduction in the number of advanced cancers (i.e., no stage shift). A follow-up study more than 20 years after the Mayo Lung Project confirmed no significant difference in lung cancer mortality. Because of its failure to reduce lung cancer mortality, chest radiograph screening for lung cancer was not recommended.
In the late 1990s, the issue of screening began to reemerge because of the ongoing debate about the validity of the findings on chest radiograph studies and in light of revolutionary developments in CT that enabled detection of pulmonary nodules smaller than 1 cm, in one breathhold, with a reduced radiation dose to the patient—low-dose CT (LDCT). The studies of lung cancer screening with CT conducted so far have been single-arm studies without a comparative group, or 1-year feasibility randomized, controlled trials. These studies showed that chest CT scans have greater sensitivity than chest radiographs for the detection of pulmonary nodules ( Fig. 1-1 ). Noncalcified nodules could be detected in as many as two thirds of the persons screened, all of whom underwent follow-up or workup to exclude malignancy, but 99% of these nodules were benign. Nodules that remained suspect for lung cancer after workup or follow-up required resection. Nevertheless, more than one third of the nodules resected were associated with benign conditions.
Despite the published 10-year survival rate of 88% for patients with stage I disease, and the increased likelihood that cancers detected by LDCT would be operable, LDCT yielded no decreases in the number of advanced lung cancers detected or in the number of deaths from lung cancers compared with predictive historical models of an unscreened population.
More recently, the National Lung Screening Trial (NLST) was launched to directly assess whether screening with LDCT is effective for early detection of lung cancer. NLST compared the effectiveness of two screening tests, LDCT and chest radiograph, on net lung cancer–specific mortality in persons who were at high risk for the development of the disease. Between September 2002 and April 2004, the trial accrued 34,614 participants, who underwent annual imaging. The trial involves follow-up questionnaires administered over 6 to 8 years and thus is still monitoring these patients; prevalence data have not yet been published. In the meantime, patients are encouraged to wait for the results of the NLST, or to be screened as part of a randomized, controlled trial, because it has not been shown that screening with LDCT is effective in reducing mortality.
Early Detection: Solitary Pulmonary Nodule
A solitary pulmonary nodule (SPN), defined as a nodule less than 3 cm in greatest dimension surrounded by lung (see Fig. 1-1 ), is a common incidental radiologic finding. Its incidence has increased with the growing use of chest CT over the past few decades and in screening studies in asymptomatic populations. Because of concern about lung cancer, further evaluation of such nodules often is suggested. The goal of imaging is to differentiate between nodules that are benign and those that are malignant, so that patients who require surgery are correctly identified; the mean postoperative mortality rate after lung cancer resection in the United States is 5%.
Chest Radiography
Evaluation of the SPN entails several steps. When a nodule is large enough to be seen on a chest radiograph, this study will be the first step in the investigation. Chest radiography is inexpensive, delivers very little radiation to the patient, and provides an image that often can be compared easily with preexisting radiographs. The initial determination is whether the nodule is indeed within the lung, because mimics of pulmonary nodules are numerous, such as rib fracture, bone island, skin lesion, or overlapping normal structures (see Fig. 1-1 ). Review of old films or old CT scans is the most cost-efficient way to assess an SPN. If no old images are available, shallow oblique images, fluoroscopy, or chest CT scan can be used.
Once the nodule has been confirmed to be within the lung, it should be assessed for features suggesting benign origin. The ability of chest radiography to discern between malignant and benign pulmonary nodules remains limited, however. Numerous studies in the 1940s and 1950s attempted to address this issue as the use of chest radiography increased exponentially. Before the advent of CT, positive preoperative diagnosis of asymptomatic SPN was rare; early exploratory thoracotomy was strongly urged for patients with these nodules. Although larger nodules are more likely to be malignant, no size criterion allows exclusion of malignancy. Two methods of distinguishing benign from malignant nodules were developed, both of which are in use today: documentation of stability of the nodule over a period of 2 years and identification of benign-appearing calcifications. Both methods are problematic: Stability was not found by robust and scientifically valid evidence to be a reliable criterion; the original data from the 1950s suggested a positive predictive value of 65% for benignity. Identifying calcifications on radiographs as benign was shown by a later study to be a subjective judgment.
Computed Tomography
In the absence of a chest radiograph from at least 2 years previously to provide a baseline for judging SPN stability, patients are referred for chest CT scan. CT is superior to chest radiography in establishing the margins and, more important, the internal characteristics of the pulmonary nodule. Spiculated margins are highly suggestive of, although not pathognomonic for, a malignant nodule ( Fig. 1-2 ). This feature can reflect the presence of fibrosis in surrounding lung parenchyma, direct infiltration of the cancer into adjacent lung parenchyma, or localized lymphangitic spread. In a study looking at 634 nodules, 50 of 53 (94%) that exhibited diffuse spiculation and 134 of 165 (81%) that showed focal spiculation were primary lung carcinomas. On the other hand, 8 of the 66 (12%) smoothly marginated, nonlobulated nodules were primary lung cancer, 6 (1%) represented a solitary metastasis, and 52 (87%) were benign. Lobulation ( Fig. 1-3 ) implies uneven growth, which often is associated with malignancy, but it is not useful in distinguishing benign from malignant nodules. Of 350 smoothly marginated lobulated nodules, 91 (26%) were primary lung cancer, 57 (16%) were metastatic disease, and 202 (58%) were benign.
With its ability to evaluate the internal characteristics of the SPN, CT revolutionized investigation of these findings. With its improved contrast resolution, elimination of overlapping structures, and slicing into thin sections, obvious calcifications can be visualized readily. For a nodule to be considered benign, obvious calcifications must be of the benign type. Characteristics of benign calcification include central, diffuse solid ( Fig. 1-4 ), and lamination ( Fig. 1-5 ) patterns and a popcorn-like appearance ( Fig. 1-6 ). Solid, central, and laminated calcification patterns typically result from a remote infection with histoplasmosis or tuberculosis (see Figs. 1-4 and 1-5 ). The popcorn-like calcification pattern is seen with hamartomas (see Fig. 1-6 ). For a nodule to be considered benign, it should display one of these four patterns of calcification and should exhibit no other features worrisome for malignancy. If calcifications are eccentric or if a nodule is bilobate, irregular, or spiculated or abuts a central bronchus, it should not be considered benign despite the presence of benign-appearing calcifications, because essentially benign calcifications can be engulfed by malignancy. In addition, because pulmonary metastatic disease from osteosarcoma or chondrosarcoma can manifest as benign-appearing calcified nodules, the calcification pattern cannot be used to differentiate benign from malignant nodules in patients with a history of one of these cancers. In such patients, benignity is established by long-term nodule stability. Another type of calcification, the sandlike, amorphous form, is seen in 6% of lung cancers imaged by CT. Such calcifications can be seen in both benign and malignant disease and thus are not be useful in diagnosis ( Fig. 1-7 ).
Although most nodules detected by CT are not obviously calcified, CT scans can objectively measure density with Hounsfield units (HU). In some previous attempts to identify subtle calcifications, not obvious to the human eye, measurements of density in HU were used to establish a threshold above which nodules were to be considered calcified and therefore benign. These attempts were based on historical studies showing that malignancies with calcifications had been identified on radiographs in less than 1% of patients. The assumption was that increased CT sensitivity would lead to identification of more benign nodules, with no false negatives, thereby reducing the number of futile thoracotomies. Subsequently, however, more than 10% of nodules evaluated as having a density higher than the established threshold of 185 HU (above which nodules should have been benign) were found to be malignant. This threshold was abandoned because it did not reliably distinguish between benign and malignant nodules.
Fat is readily recognized on CT scans. A well-demarcated nodule containing fat and having a density between −40 and −120 HU is considered benign, usually a hamartoma ( Fig. 1-8 ). A nodule consisting of fat alone or in combination with calcifications is seen in 60% of hamartomas on thin-section (using 2-mm slices) CT scans. Such a nodule, even if slow-growing (with a doubling time longer than 2 years), is considered to represent a hamartoma. Popcorn-type calcification is a typical finding in hamartoma, although other benign-type calcifications can be seen as well. A third of hamartomas do not contain calcifications or fat on CT scan and remain indeterminate nodules.
The presence of an air bronchogram within a pulmonary nodule is rare (6%) in benign nodules, but this pattern is readily identified by CT scan ( Fig. 1-9 ). Such an appearance is almost always associated with lung cancer of all cell types but is seen most commonly in adenocarcinoma (with or without bronchioloalveolar features). CT scan also can differentiate among solid nodules, those with a ground-glass appearance (in which the lung vessels can be seen through the nodule), and mixed-pattern nodules, which combine a solid portion and ground-glass portion ( Fig. 1-10 ). The malignancy rate is highest for mixed-pattern nodules (63%) and is higher for ground-glass nodules (18%) than for solid nodules (7%).
Despite the superior sensitivity of CT over radiography for detection of benign nodules by identifying fat and calcium, a majority of nodules investigated by the initial CT scan remain indeterminate. The vessels supplying tumors differ both quantitatively and qualitatively from those supplying benign growths and tend to be more “leaky.” This inherent difference in blood supply between malignant and benign nodules can be shown by changes in HU values in the pulmonary nodule after intravenous contrast injection. This method, in which the indeterminate nodule is imaged at intervals before and after intravenous contrast administration, was perfected by Swensen and associates. Absence of significant lung nodule enhancement (density of 15 HU or less) on CT is suggestive of benignity. Although the method has only 77% accuracy and 58% specificity, it does identify 98% of malignant nodules and therefore is useful in guiding follow-up or intervention.
The CT features described here will identify those patients who have nodules with benign features that do not require follow-up (benign calcifications or fat), those who would benefit from an immediate biopsy, and those who would benefit from CT monitoring of the nodule to assess its growth. The determination takes into account not only patient risk factors such as age and smoking exposure but also the CT features statistically recognized to be strongly associated with malignancy (e.g., large size, spiculation, mixed solid and ground-glass appearance). Of note, however, stability over a 2-year period is not an invariably valid criterion for benignity. In general, this criterion applies to nodules that are solid and larger than 1 cm.
Reliable detection of growth in nodules smaller than 1 cm can be difficult. For a nodule to double its volume, its diameter must increase by approximately 25%. It is difficult, even with CT, to visually detect the doubling of a 4-mm nodule, which is a change in diameter from 4 mm to 5 mm. Thus, small lung tumors can double in volume yet appear stable. Even computerized volume measurements, rather than diameter measurements, are not invariably accurate with such small nodules, which can appear to change size with differences in inspiratory effort and slice selection. Nodules with a ground-glass appearance or with a mixed solid and ground-glass pattern are detected by CT scan, not by chest radiograph, and a stability criterion for benignity, such as the 2-year stability rule used with nodules detected by chest radiograph, has not been established for such nodules on CT. In fact, such nodules, which often are detected incidentally or by screening chest CT studies, can have very long doubling times. In a screening study in Japan, the mean doubling time for ground-glass-pattern malignant nodules was 813 ± 375 days, for mixed ground-glass and solid tumors 457 ± 260 days, and for solid tumors 149 ± 125 days. In fact, 20% of the nodules in this study had doubling times exceeding 2 years, and these tended to be of the ground-glass type or mixed type. Thus, when a nodule smaller than 1 cm is monitored by CT to establish its benign nature, the follow-up period should be longer than 2 years.
Magnetic Resonance Imaging
The contrast resolution of MRI is superior to that of CT. This feature is exploited once cancer is diagnosed, because MRI is superior for evaluation of soft tissue involvement by cancer, such as in determining chest wall or nerve involvement. However, MRI does not serve effectively in early identification of lung cancer. Identifying pulmonary nodules smaller than 1 cm is hampered by the inferior spatial resolution of MRI, which is particularly poor in the lungs, as a consequence of characteristics both of the lungs themselves, such as low proton density and numerous air-tissue interfaces, and of the examination, such as motion artifacts from respiratory and cardiac motion. Dynamic contrast-enhanced MRI has been shown in small studies to have sensitivity rates for differentiation of malignant from benign SPNs that were comparable with those obtained with dynamic contrast-enhanced CT, but the nodules investigated usually were larger than the incidental nodules discovered by CT.
Positron Emission Tomography
PET imaging with F-fluorodeoxyglucose (FDG) has emerged as an additional tool for evaluation of the SPN. FDG-PET is a physiologic imaging modality, with poor spatial resolution in comparison with morphologic imaging modalities such as chest CT or radiograph. This technique assesses use of glucose by different body structures based on the preferential uptake of F-FDG by metabolically active tissue. Because many cancers, including non–small cell lung cancer (NSCLC), have a higher metabolic rate than that of surrounding normal tissue, they accumulate F-FDG more intensely and therefore appear “hot” on PET images. For nodules that are indeterminate on CT investigation, PET scan can help identify patients who may benefit from immediate biopsy. Initial studies showed that FDG-PET was effective in the differentiation of benign from malignant pulmonary lesions, and several early reports suggested that PET examinations reduce the number of patients with indeterminate nodules who undergo unnecessary thoracotomy, with overall sensitivity, specificity, and accuracy estimated to be 96%, 88%, and 94%, respectively. PET is neither uniformly specific nor sensitive, however, particularly if the abnormality is small. Nodules smaller than 1 cm are not measured accurately and sometimes fall below the resolution of the PET scan.
Although the combination of PET with a CT scan, or integrated PET-CT , has been shown to provide significantly greater specificity than that for either study alone, the quantification of FDG uptake with use of CT for attenuation correction can introduce an artifact related to different breathing states in the CT and PET scans. FDG uptake in nodules, particularly those in the lower lungs, which suffer greater motion from the breathing cycle, will then erroneously appear to be lower than is actually the case.
Cell type also influences FDG uptake. Indolent cancers, such as carcinoid tumors, well-differentiated adenocarcinomas, or bronchioloalveolar cell carcinoma (BAC), demonstrate less FDG activity than that seen in other NSCLCs and in some cases show no increased FDG activity. The typical features of some of these cancers, such as proximity to a bronchus as is common with carcinoid tumors or the consolidative or ground-glass nodule in some BACs, are taken into account in interpreting the results of the PET-CT scan. The negative PET result thus serves as a tool, not a definite marker of benignity. If biopsy is deferred, the SPN with the negative PET result is monitored with serial chest CT scans for growth of the lesion. The data gathered thus far indicate that PET-negative nodules are indolent cancers; accordingly, this approach should not adversely affect patient outcome.
The positive predictive value of PET in most patients is high (90% if the patient is older than 60 years). False-positive studies of the primary lesion (a positive FDG-PET result with a lesion that proves to be benign) have been reported with infectious and inflammatory processes such as tuberculosis, histoplasmosis, and rheumatoid nodules. Lesions with increased FDG uptake, however, should be considered malignant until proven otherwise and should be managed accordingly.
Imaging of Lung Cancer Subtypes
Imaging cannot replace histologic sampling of lung masses, but certain subtypes of lung cancer can manifest with typical imaging features.
Squamous cell carcinoma typically originates centrally, so the presenting manifestation frequently is postobstructive pneumonia or atelectasis, which is readily identified on the chest radiograph ( Fig. 1-11 ). Less common manifestations are mucoid impaction, bronchiectasis, and hyperinflation. Approximately one third of squamous cell carcinomas arise beyond the segmental bronchi. Squamous cell carcinomas are more likely to cavitate than the other histologic subtypes of lung cancer. Cavitation occurs in 10% to 30% of these cancers and is more common in large peripheral masses and poorly differentiated tumors. Because most squamous cell carcinomas grow slowly and become symptomatic because of their central location, extrathoracic metastases are encountered less commonly in imaging at presentation.
Adenocarcinomas typically manifest as peripheral SPNs (see Fig. 1-9 ). Historically, nodules have been described as typically having soft tissue attenuation and an irregular or spiculated margin. With the expanding use of CT and screening studies, however, an increasing number of adenocarcinomas manifest as nodules with a ground-glass appearance on CT or with mixed ground-glass and solid components (see Fig. 1-10 ). A correlation has been found between these CT appearances and the classification proposed by Noguchi and coworkers, whereby small (2 cm or less in greatest dimension) peripheral adenocarcinomas are classified into six types based on tumor growth patterns: type A, localized BAC; type B, localized BAC with foci of structural collapse of alveoli; type C, localized BAC with active fibroblastic proliferation; type D, poorly differentiated adenocarcinoma; type E, tubular adenocarcinoma; and type F, papillary adenocarcinoma with a compressive growth pattern. Ground-glass attenuation of nodular opacities has been reported to be more frequent in types A to C than in types D to F, whereas soft tissue attenuation is more frequent in types B to F. The soft tissue attenuation component tends to be absent or less than a third of the opacity with type A and greater in extent (more than two thirds) in types D to F. Mixed nodules, with both ground-glass and solid components, have a higher likelihood of being invasive and of higher stage than nodules with a pure ground-glass appearance.
Although BAC is known to manifest with the unusual appearance of consolidation, this presentation is seen in only 30% of the cases; the rest of these tumors manifest as SPNs (43%) or multiple nodules (30%). The SPNs are usually peripherally located and can remain stable in size for many years, with doubling times greater than 2 years. They can be of the ground-glass type or mixed type, with cystic changes or cavitation occurring rarely, in up to 7%. When a nodule exhibits multiple small, focal low-attenuation regions (pseudocavitation) or air bronchograms, the diagnosis of BAC should be suspected. On PET-CT scans, BAC can show low FDG activity, lower than expected for malignancy.
Large cell carcinoma usually manifests as a peripheral, poorly marginated large mass (larger than 7 cm in greatest dimension). Although growth typically is rapid, cavitation is uncommon.
The most common presentation of carcinoid tumors is that of a central endobronchial mass, with or without atelectasis or consolidation, or, less commonly, a well-demarcated pulmonary nodule. The tumors usually are less than 3 cm in diameter ( Fig. 1-12 ), although occasionally they may be as large as 10 cm. Calcification is seen in 25% of carcinoids by CT. Carcinoids can show low FDG uptake on PET-CT studies, lower than expected for malignancies.
The primary tumor of small cell lung cancer (SCLC) typically is small, in a central location, and associated with marked hilar and mediastinal adenopathy, frequently with engulfment of the primary lesion until it is no longer identifiable ( Fig. 1-13 ). With the increased use of CT and screening CT scans, the number of SCLCs encountered as early, small peripheral SPNs without intrathoracic adenopathy has increased. In the literature, detection of such early disease was reported in only 5% of the cases.
Staging
Non–Small Cell Lung Cancer
Accurate staging of lung cancer is important in determining disease management and prognosis. The primary goal of radiologic staging is to distinguish disease that is potentially resectable (stages I to IIIA) from nonresectable disease (stages IIIB and IV). The current TNM staging system assesses the primary tumor (T), spread into local lymph nodes (N), and distant spread, or metastasis (M). This system originally was designed for conventional anatomic assessment and does not take into account information from FDG-PET scans, although the PET data currently are being integrated into this staging system, and the use of this modality is described in this section. The TNM system proposed in 1997 is in current use ( Tables 1-1 and 1-2 ). A proposed revision to this TNM staging system was published recently and has been implemented in some academic centers, but it has not yet been fully implemented in all clinical practices ( Tables 1-3 and 1-4 ).
| Primary Tumor (T) | |
| TX | Primary tumor cannot be assessed or Tumor proven by the presence of malignant cells in sputum or bronchial washings but not visualized by imaging or bronchoscopy |
| T0 | No evidence of primary tumor |
| Tis | Carcinoma in situ |
| T1 | Tumor ≤ 3 cm in greatest dimension, surrounded by lung or visceral pleura, without bronchoscopic evidence of invasion more proximal than the lobar bronchus * (i.e., not in the main bronchus) |
| T2 |
|
| T3 | Tumor of any size that directly invades any of the following: chest wall (including superior sulcus tumors), diaphragm, mediastinal pleura, parietal pericardium or Tumor in the main bronchus < 2 cm distal to the carina, but without involvement of the carina, or tumor associated with atelectasis or obstructive pneumonitis of the entire lung |
| T4 | Tumor of any size that invades any of the following: mediastinum, heart, great vessels, trachea, esophagus, vertebral body, carina or Tumor with a malignant pleural or pericardial effusion, † or with satellite tumor nodule(s) within the ipsilateral primary tumor lobe of the lung |
| Regional Lymph Nodes (N) | |
| NX | Regional lymph nodes cannot be assessed |
| N0 | No regional lymph node metastasis |
| N1 | Metastasis to ipsilateral peribronchial and/or ipsilateral hilar lymph nodes, and intrapulmonary nodes involved by direct extension of the primary tumor |
| N2 | Metastasis to ipsilateral mediastinal and/or subcarinal lymph node(s) |
| N3 | Metastasis to contralateral mediastinal, contralateral hilar, ipsilateral or contralateral scalene, or supraclavicular lymph node(s) |
| Distant Metastasis (M) | |
| MX | Presence of distant metastasis cannot be assessed |
| M0 | No distant metastasis |
| M1 | Distant metastasis present ‡ |
* The uncommon superficial tumor of any size with its invasive component limited to the bronchial wall, which may extend proximal to the main bronchus, also is classified T1.
† Most pleural effusions associated with lung cancer are due to tumor. In a few patients, however, abundant cytopathologic evidence indicates that the effusion is not related to the tumor; in such cases, the effusion should be excluded as a staging element and the patient’s disease should be staged T1, T2, or T3. Disease associated with pericardial effusion is classified according to the same rules.
‡ Separate metastatic tumor nodule(s) in the ipsilateral non–primary tumor lobe(s) of the lung also are classified M1.
| Stage | TNM Subset |
|---|---|
| 0 | Carcinoma in situ |
| IA | T1N0M0 |
| IB | T2N0M0 |
| IIA | T1N1M0 |
| IIB | T2N1M0 |
| T3N0M0 | |
| IIIA | T3N1M0 |
| T1N2M0 | |
| T2N2M0 | |
| T3N2M0 | |
| IIIB | T4N0M0 |
| T4N1M0 | |
| T4N2M0 | |
| T1N3M0 | |
| T2N3M0 | |
| T3N3M0 | |
| T4N3M0 | |
| IV | Any T any N M1 |
* Staging is not relevant for occult carcinoma, designated TXN0M0.
| T (Primary Tumor) | |
| TX | Primary tumor cannot be assessed or Tumor proven by the presence of malignant cells in sputum or bronchial washings but not visualized by imaging or bronchoscopy |
| T0 | No evidence of primary tumor |
| Tis | Carcinoma in situ |
| T1 | Tumor ≤ 3 cm in greatest dimension, surrounded by lung or visceral pleura, without bronchoscopic evidence of invasion more proximal than the lobar bronchus (i.e., not in the main bronchus) * |
| T1a | Tumor ≤ 2 cm in greatest dimension |
| T1b | Tumor > 2 cm, but ≤3 cm in greatest dimension |
| T2 | Tumor > 3 cm, but ≤7 cm in greatest dimension or Tumor with any of the following features (T2 tumors with these features are classified T2a if ≤5 cm in size)
|
| T2a | Tumor > 3 cm, but ≤5 cm in greatest dimension |
| T2b | Tumor > 5 cm, but ≤7 cm in greatest dimension |
| T3 | Tumor > 7 cm in greatest dimension or Tumor that directly invades any of the following: chest wall (including superior sulcus tumors), diaphragm, phrenic nerve, mediastinal pleura, parietal pericardium or Tumor in the main bronchus < 2 cm distal to the carina but without involvement of the carina, or associated with atelectasis or obstructive pneumonitis of the entire lung or separate tumor nodule(s) in the same lobe |
| T4 |
|
| N (Regional Lymph Nodes) | |
| NX | Regional lymph nodes cannot be assessed |
| N0 | No regional lymph node metastasis |
| N1 | Metastasis in ipsilateral peribronchial and/or ipsilateral hilar lymph nodes and intrapulmonary nodes, including involvement by direct extension |
| N2 | Metastasis in ipsilateral mediastinal and/or subcarinal lymph node(s) |
| N3 | Metastasis in contralateral mediastinal, contralateral hilar, ipsilateral or contralateral scalene, or supraclavicular lymph node(s) |
| M (Distant Metastasis) | |
| MX | Distant metastasis cannot be assessed |
| M0 | No distant metastasis |
| M1 | Distant metastasis |
| M1a | Separate tumor nodule(s) in a contralateral lobe; tumor with pleural nodules or malignant pleural (or pericardial) effusion † |
| M1b | Distant metastasis |
* The uncommon superficial spreading tumor of any size with its invasive component limited to the bronchial wall, which may extend proximally to the main bronchus, is also classified as T1.
† Most pleural (and pericardial) effusions with lung cancer are due to tumor. In a few patients, however, findings on multiple cytopathologic examinations of pleural (pericardial) fluid are negative for tumor, and the fluid is nonbloody and is not an exudate. Where these elements and clinical judgment dictate that the effusion is not related to the tumor, the effusion should be excluded as a staging element and the patient’s disease should be classified as T1, T2, T3, or T4.
| Previous T/M Descriptor | Proposed T/M | N0 | N1 | N2 | N3 |
|---|---|---|---|---|---|
| T1 (≤2 cm) | T1a | IA | IIA | IIIA | IIIB |
| T1 (>2–3 cm) | T1b | IA | IIA | IIIA | IIIB |
| T2 (≤5 cm) | T2a | IB | IIA | IIIA | IIIB |
| T2 (>5–7 cm) | T2b | IIA | IIB | IIIA | IIIB |
| T2 (>7 cm) | T3 | IIB | IIIA | IIIA | IIIB |
| T3 invasion | IIB | IIIA | IIIA | IIIB | |
| T4 (same lobe nodules) | IIB | IIIA | IIIA | IIIB | |
| T4 (extension) | T4 | IIIA | IIIA | IIIB | IIIB |
| M1 (ipsilateral lung) | IIIA | IIIA | IIIB | IIIB | |
| T4 (pleural effusion) | M1a | IV | IV | IV | IV |
| M1 (contralateral lung) | IV | IV | IV | IV | |
| M1 (distant) | M1b | IV | IV | IV | IV |
Although consensus on the optimal imaging modality for the staging of lung cancer is elusive, evidence-based guidelines were published by American Society of Clinical Oncology (ASCO) in 2004. Initial evaluation is recommended to include a chest radiograph and a contrast-enhanced chest CT scan that encompasses the adrenal glands and liver. A PET scan is recommended for further evaluation in cases in which CT provides no evidence of metastatic disease. This recommendation is based on the fact that FDG-PET imaging improves nodal and distant metastatic staging and frequently alters staging to a degree that changes management. The use of integrated PET-CT scanners has further enhanced the accuracy of staging of NSCLC, because these two studies, when performed together, complement each other by overcoming the lack of spatial resolution inherent in the PET scan and the lack of physiologic information inherent in the CT scan. These studies, in combination with clinical and laboratory findings, are then used to determine the necessity of additional imaging studies, as discussed later in this section.
Primary Tumor (T Status)
The T status is defined by the primary cancer’s size, location, and invasion into surrounding structures. Because of the inferior spatial resolution of the PET scan, it is not used for T staging, which takes into account morphologic features alone. This restriction holds despite evidence that the amount of FDG uptake does correlate with prognosis, and that patients whose primary tumor has higher FDG avidity, even if early-stage, have a shorter survival. The proposed staging system includes changes to T status: T1 has been subcategorized as T1a, for tumors 2 cm or less in greatest dimension, or T1b, for tumors more than 2 to 3 cm or less in greatest dimension; T2 has been subcategorized as T2a, for tumors more than 3 to 5 cm or less in size, or T2, for tumors associated with certain other factors (see Table 1-1 ) and 5 cm or less in size, or T2b, for tumors more than 5 to 7 cm or less in size; T2 tumors larger than 7 cm have been reclassified as T3; T4 tumors with additional nodule(s) in the same lobe have been reclassified as T3; M1 tumors with additional nodule(s) in the same lung have been reclassified as T4; and T4 pleural dissemination has been reclassified as M1.
CT is the best overall imaging modality for determining the T stage, which usually is size-dependent. CT can readily identify features of more advanced T stage, such as gross chest wall involvement with rib destruction or bulging chest wall abnormality. CT is inaccurate, however, for identifying subtle chest wall involvement, such as involvement of the parietal pleura rather than tumor merely abutting the structure. In one study, the sensitivity of CT in distinguishing T3–T4 tumors from T0–T2 tumors was 63% and specificity was 84%. Some subtle CT criteria suggestive of chest wall invasion are obliteration of the extrapleural fat plane, tumor-pleura contact extent greater than 3 cm in length, higher ratio of tumor-pleura contact extent to tumor height, and formation of an obtuse angle between tumor and pleura. Despite the superior contrast resolution of MRI, its accuracy in identifying chest wall invasion is insufficient and similar to that of CT. Although ultrasound imaging has a very limited role in the evaluation of patients with NSCLC, because the air within the lungs interferes with sound wave transmission, it affords better soft tissue resolution than that obtained with CT and has the advantage of providing real-time imaging throughout the respiratory cycle. For detection of chest wall involvement, ultrasound imaging is superior to CT, with a sensitivity of 89% (compared with 40% for CT) and similar specificity, approaching 100%.
For detection of direct mediastinal involvement, CT and MRI findings suggestive of subtle invasion of the mediastinum are tumor contact extent greater than 3 cm with mediastinum, angle of contact with aorta greater than 90 degrees, and lack of mediastinal fat between the mass and mediastinal structures. Although MRI was found in one study to be superior to CT in identifying direct mediastinal involvement, accuracy of both modalities in assessment of mediastinal involvement was disappointing, with a sensitivity of 55% for CT and 64% for MRI.
The superb soft tissue contrast resolution and multiplanar capability of MRI are ideally used for evaluation of superior sulcus tumors. In a retrospective study of 143 patients with superior sulcus tumors, longer survival was associated with surgery in the absence of nodal disease. An absolute contraindication to surgery is tumor invasion of the brachial plexus roots or trunks above the level of the T1 nerve root. The similarity of the brachial plexus to its surrounding structures, its superior-inferior orientation, and its small size make it almost impossible to evaluate accurately in the axial plane. It is, however, readily identified in the sagittal plane by MRI. Although NSCLC T4 lesions, such as those involving the vertebral body, generally are considered unresectable, superior sulcus tumors that involve less than 50% of the vertebral body may be resectable, often with the aid of the neurosurgical team. At the time of imaging, MRI can determine whether the carotid artery and the vertebral artery are involved by tumor; such involvement is a relative contraindication to surgery. MRI also can determine if the contralateral vessels are severely affected by atherosclerotic disease, in which case surgery may not be an option. In addition to determining resectability, imaging plays a vital role in selecting the most appropriate surgical approach. Posteriorly located tumors are amenable to resection through a posterolateral incision, but tumors that involve the trunks of the brachial plexus or the subclavian vessels usually require an anterior approach.
Over the years, different MRI sequences have been developed to overcome flow artifacts and to improve vascular and cardiac images in motion. MRI is used to assess whether the tumor directly involves the heart and, if it does, to what extent. It has been shown that in a select group of patients with T4 lesions involving the heart ( Fig. 1-14 ), a trimodality approach (chemotherapy and concomitant radiotherapy plus surgery) improved survival. Patients who underwent tumor resection had a significantly better 5-year survival rate than that in patients who did not: 38% versus 5.6%.
In the current staging system, disease presenting as a malignant pleural effusion falls into the T category of staging, as T4 disease; in the proposed new system, such cases should fall into the M category of disease. Unfortunately, it is frequently difficult to establish the diagnosis of a malignant pleural effusion, because fluid obtained at thoracentesis is positive for malignancy in only 66% of patients, and the pleura does not always show nodularity on CT imaging. Imaging with PET is helpful, but studies on the accuracy of PET in establishing the diagnosis of a malignant effusion are few, with reported sensitivities of 92% to 100%, specificities of 67% to 71%, negative predictive values of 100%, and positive predictive values of 63% to 79%. PET scans for pleural malignancy should be interpreted with caution and in conjunction with the CT scan, as inflammation after talc pleurodesis can last for years, and show increased FDG uptake in the absence of malignant cells ( Fig. 1-15 ). A negative PET result can be useful, however, in confirming the absence of pleural metastatic disease, particularly when the results of thoracentesis are also negative.
Nodal Disease (N Status)
The proposed new staging system includes no changes in nodal staging (see Tables 1-1 and 1-3 ). The role of the chest radiograph in the nodal staging of NSCLC is usually minor, as it is usually insensitive to mild and modest nodal enlargement. Bulky bilateral adenopathy dictates a diagnosis of stage IIIB. If the patient is too ill or is unwilling to undergo treatment, this should suffice for staging. For the majority of patients, however, more accurate staging is needed.
CT scanning is routinely used for noninvasive staging of the lymph nodes. The sole criterion for differentiating benign from malignant lymph nodes, by CT or by MRI, is size. The most widely used criterion for identifying a malignant lymph node is a short axis diameter greater than 1 cm. This criterion was chosen as a fine balance between sensitivity and specificity, in order to minimize false-negative evaluations. Numerous studies have looked at the performance of CT in distinguishing benign and malignant lymph nodes in patients with NSCLC. A large study that pooled lymph node data for 5111 patients from 43 different studies found that the sensitivity of CT for detecting metastases in the mediastinal lymph nodes was 51% and the specificity 86%. Similarly, two meta-analyses showed sensitivity rates of 61% to 64% and specificities of 74% to 79%. The accuracies of CT and MRI in detecting nodal metastases are similar: the accuracy of CT ranges from 56% to 82% and that of MRI from 50% to 82%. Such poor performance is due to the fact that normal-sized lymph nodes may harbor tumor and that nodal enlargement may be a response to a benign reactive process. Recent attempts to abandon the size criteria for malignancy in favor of MRI examination of internal characteristics of the lymph node, such as high signal intensity, eccentric cortical thickening, or obliterated fatty hilum, have shown similar disappointing results, with accuracy rates of 70% to 73%.
The accuracy of PET is superior to that of CT in nodal staging, but the results should be interpreted with caution and in conjunction with CT findings. Non-neoplastic inflammatory processes also show increased FDG activity. As with the pulmonary nodule, PET is less accurate in the evaluation of lymph nodes smaller than 10 mm. In a pooled analysis of multiple studies in which a total of 2865 patients were evaluated, the sensitivity and specificity of PET for identifying metastatic lymph nodes were 74% and 85%, respectively. In a meta-analysis of 17 studies comprising 833 patients, the overall sensitivity of PET for detecting nodal metastases was 83% and the specificity was 92%, whereas the sensitivity and specificity of chest CT were 59% and 78%, respectively. Integrated PET-CT improves nodal staging over that achieved with PET alone. In the presence of enlarged lymph nodes, PET and PET-CT become less specific and less accurate but more sensitive in detecting nodal metastatic spread. In one meta-analysis, the median sensitivity and specificity of PET scans were 100% and 78%, respectively, in patients with enlarged lymph nodes. The reduced specificity in the presence of enlarged lymph nodes means that almost one fourth of patients diagnosed with metastatic lymph nodes actually had no nodal metastasis but rather a reactive or inflammatory lymphadenopathy.
Because of these limitations, ASCO recommends that a confirmatory biopsy be performed in cases of FDG-avid mediastinal lymph nodes, so that patients with operable disease will not be denied curative surgery. A PET scan is justified even when the initial chest CT scan confirms highly suspect mediastinal lymph nodes. The PET scan may influence the site of biopsy by identifying a previously unsuspected location of metastatic disease (which may upstage the disease) or a location that is safer to biopsy. In patients whose mediastinal lymph nodes are smaller than 1 cm, approximately 20% will show false-negative findings on PET scans; in a meta-analysis, the sensitivity and specificity of PET were 82% and 93%, respectively. Although PET demonstrates all lymph node stations, whereas mediastinoscopic or transbronchial lymph node biopsy is unable to sample all lymph node stations, biopsy remains the most accurate preoperative measure for identifying occult metastatic disease in mediastinal lymph nodes smaller than 1 cm.
Distant Metastases (M Status)
The purpose of staging is to detect metastatic disease, particularly at the common metastatic sites for NSCLC—the adrenal glands, liver, brain, and bone—with the goal of preventing nontherapeutic thoracotomy. In the recently proposed revision to the TNM staging system, M status has been divided into M1a for metastases within the thoracic cavity and M1b for extrathoracic metastatic disease. The M1a category includes malignant pleural effusions and malignant pleural nodules, previously designated T4, and metastatic pulmonary nodules to the contralateral lung.
Beyond the initial chest CT scan, little consensus has emerged regarding the optimal noninvasive staging for NSCLC. When patients are found to have early disease (stage I or II) by the initial chest CT examination, with no clinical symptomatology, the yield for imaging for additional metatatic disease is low. Although some proponents advise further extrathoracic staging for tumors whose histologic type is associated with a higher likelihood of extrathoracic metastasis at the time of presentation, such as adenocarcinoma or large cell carcinoma, this approach was not found to be productive in a large series of patients with early-stage lung cancer. In addition, results of each imaging modality should be interpreted with caution, because biopsy of each suspected metastatic site is not feasible, and results often rely on follow-up or comparison with other imaging modalities. In a study examining biopsy specimens of normal-appearing adrenal glands in patients with NSCLC staged by chest CT scan, 12% of the glands were found to harbor metastatic disease. A more recent study that compared autopsy results with findings on CT scan of the adrenal glands obtained within 90 days before death showed that CT detected only 20% of the metastatic adrenal glands. This low sensitivity was considered to be due to the lack of substantial structural change in many of these adrenal glands.
On the basis of these findings, recent guidelines of the American College of Chest Physicians and the latest ASCO recommendations recommend PET scanning for staging but also further imaging in accordance with to symptomatology or for abnormal lesions that remain indeterminate after the initial investigations with PET and CT. Except in the brain, PET has a higher sensitivity and specificity than CT or bone scan in detecting metastatic disease. In a study of 303 patients, the sensitivity and specificity for detection of M1 disease by PET were 83% and 90%, respectively. An average of 15% of patients have unexpected distant metastases detected by PET, and in 20% of patients, findings on PET imaging preclude nontherapeutic thoracotomy. PET scanning has the advantage of imaging the entire body with one examination and assessing areas not covered by conventional imaging, such as the skin, muscles, and pelvis, for detection of unusual metastatic foci.
The adrenal glands are the most common site of metastasis in patients with NSCLC, and adrenal metastasis can occur as an isolated site in as many as 6% of patients. Adrenal masses are found in as many as 20% of patients at initial presentation, yet a majority are benign. CT or MRI features suggesting that an adrenal nodule is malignant include size greater than 3 cm, poorly defined margins, an irregularly enhancing rim, invasion of adjacent structures, and high signal intensity on T2-weighted MRI sequences. When CT shows an adrenal nodule with density measured as 10 HU or less, a confident diagnosis of adrenal adenoma is made. This finding has 98% sensitivity but only 71% specificity, because 30% of adenomas do not contain a sufficient amount of lipid to be measurable by CT. In these cases, an effective choice is MRI using chemical shift analysis to differentiate a benign from a malignant nodule. Results of chemical shift analysis (with MRI) and HU measurement (with CT) can include errors when the adrenal gland nodule is small.
PET imaging is excellent in establishing that an adrenal nodule is benign. Although early studies suggested that the accuracy of PET in determining the nature of an adrenal mass was 100%, further experience with PET scanning has shown that, although the sensitivity and specificity are high at 100% and 80% to 90%, respectively, increased FDG uptake can be seen in adenomas. Greater accuracy is obtained when an adrenal nodule is found to have greater FDG activity than that of the liver, rather than using a specific standardized uptake value (SUV) as a threshold ( Fig. 1-16 ). Because uptake can be high in adenomas, ASCO recommends that an isolated adrenal mass on an ultrasound image, CT scan, or FDG-PET scan be biopsied to rule out metastatic disease if the lesion is otherwise considered to be potentially resectable.
Although NSCLC frequently metastasizes to the liver, it is unusual for the liver to be an isolated site of disease, particularly in the absence of metastatic disease to regional lymph nodes. In most cases, therefore, the finding of liver metastases does not significantly alter management of the disease. One meta-analysis found that only 3% of asymptomatic patients with NSCLC will have liver metastases on CT scan. Although PET has been found to detect liver metastases with accuracy rates ranging between 92% and 100%, with only rare false-positive findings, the data from available studies are limited and were not compared with results of systematic biopsies or state-of-the-art liver imaging. When a liver lesion is suspected to represent metastatic disease by any imaging modality, it needs to be confirmed with biopsy if the disease is considered to be potentially resectable.
Routine evaluation for brain metastases in asymptomatic patients presenting with newly diagnosed NSCLC remains controversial and is not universally recommended by ASCO. When brain CT is performed in asymptomatic patients staged for NSCLC, the median prevalence of brain metastases is 3% (range, 0% to 21%), with a median predictive value of a negative clinical evaluation of 97%, whereas when brain CT is performed in both symptomatic and asymptomatic patients, the prevalence of brain metastases is 14% (range, 6% to 32%). Asymptomatic brain metastases are more commonly found in patients with advanced intrathoracic disease. The detection rate for patients with stage I or II disease is 4% with imaging by CT or MRI, whereas a detection rate of 11.4% has been reported for those with stage III disease. Although MRI can detect smaller and more numerous brain metastases, no studies have been conducted showing that MRI is better than CT at identifying patients with metastases from NSCLC. Consequently, ASCO recommends that either CT or MRI is acceptable for imaging for brain metastases. Either study should be performed in patients who have neurologic signs or symptoms, as well as in asymptomatic patients with stage III disease who are being considered for aggressive local therapy such as thoracotomy or irradiation. PET is not recommended for imaging of brain metastases: PET performs poorly in the evaluation of brain metastases because FDG avidly accumulates in the gray matter, limiting detectability of metastatic disease, which usually occurs in the same region. Sensitivity for detection of brain metastases by PET can be as low as 60%.
Although patients with skeletal metastases usually are symptomatic or have laboratory abnormalities indicating bone metastases, 27% of asymptomatic patients in one study were found to have skeletal metastases. However, false-positive abnormalities on technetium-99m methylene diphosphonate bone scintigraphy are numerous, owing to the frequency of degenerative and traumatic skeletal changes. PET scanning is superior to bone scintigraphy in identifying skeletal metastases: PET not only is able to view marrow metastases that typically are not detected by bone scintigraphy but also yields few false-positive results. The specificity, sensitivity, negative predictive value, positive predictive value, and accuracy of PET scanning in the assessment of bone metastases all exceed 90%. Accordingly, ASCO’s position is that bone scintigraphy is optional in patients who have evidence of bone metastases by PET scanning, unless suggestive symptomatology is noted in regions not imaged by PET. Because of the possibility of false-positive uptake with both PET and bone scintigraphy, patients who are operative candidates are required to have histologic confirmation or corroboration by morphologic imaging (plain radiography, CT, or MRI) of a lesion that will increase the stage of their disease.
To summarize staging for NSCLC, imaging with a chest CT scan that includes the adrenal glands is routine. If disease does not appear to be metastatic, further staging with PET or PET-CT is recommended. Biopsy of suspect mediastinal lymph nodes (i.e., those larger than 1 cm or with increased FDG activity) is needed for confirmation of nodal disease. Patients with locally advanced disease who are to undergo aggressive therapy (surgery or irradiation) should undergo dedicated brain imaging (MRI or contrast-enhanced CT) even if they are asymptomatic. Additional imaging, such as brain imaging for early disease or dedicated bone imaging (plain film, scintigraphy, or MRI) is performed if the patient is symptomatic, or to clarify equivocal imaging findings on the initial PET and CT studies. When a metastatic focus is found that would change clinical management, such as one metastatic lesion in a patient whose disease is otherwise resectable, it should be verified with biopsy.
Small Cell Lung Cancer
Compared with imaging studies of NSCLC, studies on the usefulness of imaging in the staging of SCLC are few. This lack may be related to the dismal prognosis for the disease, to the fact that the great majority of patients are treated nonsurgically, or to the simplified method of dichotomous staging developed by the Veterans Administration Lung Cancer Study Group. According to this method, limited disease includes tumors confined to the hemithorax of origin, the mediastinum, and/ or the supraclavicular lymph nodes. In extensive disease, tumor spreads beyond those limited sites. Most patients presenting with SCLC have disseminated disease at initial staging. Since the common sites of metastatic disease are the liver, bone, bone marrow, brain, and retroperitoneal lymph nodes, many of the patients with metastatic disease are identified at the initial staging chest CT scan, but there is no concensus as to the routine use of imaging modalities in this disease.
Multiple studies are routine, including bone marrow aspiration, brain MRI, CT of the chest and abdomen, and bone scintigraphy. Attempts have been made to image the entire body with one imaging modality and eliminate the multiplicity of studies. Although this can be done with MRI, it has not gained popularity. Lately, attempts have been made to popularize staging with PET or PET-CT, but this was not embraced in the management guidelines issued recently by the American College of Chest Physicians, mainly because a majority of these studies investigated fewer than 50 patients and lacked reference standards to verify staging accuracy. As in NSCLC, PET-CT is more accurate than chest CT staging alone and is inferior to conventional imaging when assessing for brain metastases. Most recent reports suggest that staging with PET entails a change in management in 8% to 16% of patients with SCLC.
Extensive assessment for bone metastases (i.e., bone marrow aspiration, bone scintigraphy, and MRI) is not performed for asymptomatic patients with limited disease. It is usually reserved for patients with extensive disease because isolated bone metastases and bone marrow metastases are not common.
Brain metastases, on the other hand, are common at presentation. Because they are seen at presentation in as many as 24% of asymptomatic patients who undergo contrast-enhanced brain MRI, such imaging has been advocated by many experts as part of routine staging.
Liver metastases and retroperitoneal lymph node metastases usually are asymptomatic yet are common at presentation of SCLC. Staging should therefore include the entire liver, and imaging for this purpose should be performed with intravenous contrast, by either CT or MRI.
Follow-up Evaluation
Assessing Response to Chemotherapy
Accurately assessing a cancer’s response to chemotherapy is important clinically, for individual patients (both surgical and nonsurgical candidates) and for trials assessing the efficacy of novel anticancer therapies. In patients with potentially resectable lung cancer receiving chemoradiotherapy, evaluating the response of the tumor to therapy is important in assessing the efficacy of treatment and predicting the long-term prognosis. This information is of potential value in helping to determine which patients will benefit most from surgery and which patients may require additional nonsurgical treatment. In patients who are not surgical candidates, or whose disease is not responding to therapy, potentially toxic and expensive chemotherapeutic drugs may be changed or discontinued. Small differences in patients’ response rates can affect the outcome of phase I and II clinical trials, which may dictate which new drugs are introduced to the market. Accordingly, uniform, reproducible, and accurate response criteria are essential.
Traditionally, response to therapy has been assessed by measuring tumor volume. Because calculation of tumor volume is cumbersome, simplified methods have been applied over the years that are correct for spherical tumors. Response criteria proposed in 1979 after a meeting on the Standardization of Reporting Results of Cancer Treatment were widely accepted. These criteria, known as the World Health Organization (WHO) criteria for reporting the results of cancer treatment, are based largely on tumor measurements in two dimensions—the two longest perpendicular diameters in the axial plane that are perpendicular to each other. In 1994, the WHO criteria were reviewed, and revised guidelines known as Response Evaluation Criteria in Solid Tumors (RECIST) were proposed. These guidelines recommended determination of treatment response using a single measurement of the largest tumor diameter in the axial plane.
Measurements using the WHO and RECIST criteria can be inaccurate for nonspherical tumors and for tumors with indistinct margins. When these criteria were proposed, a generalized assumption was that they would allow accuracy and reproducibility of measurements performed by different readers. The accuracy of the criteria has been questioned, however, because a recent study demonstrated great interobserver variability in the measurement of tumors, potentially leading to incorrect interpretations of tumor response. Consistency in measured diameters was improved when the same reader performed serial tumor measurements—a protocol that can be implemented in clinical trials but is not always feasible in routine daily practice.
PET imaging after the initiation of chemotherapy or radiotherapy can assess the response of the primary tumor to treatment by detecting a reduction in metabolic activity of the primary mass, a favorable prognostic indicator of survival for both patients with NSCLC and those with SCLC. In a prospective study in 60 patients with stage III NSCLC who underwent neoadjuvant chemoradiotherapy before surgical resection, a restaging PET study performed 2 weeks after induction therapy was able to predict the pathologic response in the primary tumor, determined at subsequent surgery, with a sensitivity of 86% and a specificity of 81%. In another prospective study in 57 patients with locally advanced NSCLC who underwent restaging PET imaging after only one cycle of platinum-based chemotherapy, a fall in SUV max of 20% or greater in the primary tumor was an independent predictor of long-term survival. Median survival duration was 252 days in responders but only 151 days in nonresponders. Because SUV is only a semiquantitive measurement affected by multiple technical factors, small changes in SUV are considered insignificant.
In 1999, after reviewing FDG-PET oncology studies, the European Organization for Research and Treatment of Cancer PET study group published universal guidelines for determination of tumor response that have been applied to studies. These have been implemented in daily practice and some study designs, but have not yet been implemented into the RECIST guidelines. These guidelines define complete metabolic response as complete resolution of FDG activity in the tumor; partial metabolic response as a decrease of SUV by 15% to 25% after one chemotherapy cycle, or a decrease of greater than 25% after more than one chemotherapy cycle; stable metabolic disease as an increase in SUV of less than 25% or decrease of SUV of less than 15%; and progressive disease as an increase in SUV by more than 25%.
Some issues with PET imaging in the evaluation of tumor response remain unresolved. The ideal timing of the study has not yet been demonstrated, because inflammatory response from therapy, such as associated with radiation therapy, increases FDG activity as well. FDG-PET does not appear to offer any advantages over CT for lymph node staging or for predicting the pathologic response after neoadjuvant treatment of NSCLC. In patients who have received neoadjuvant therapy before planned surgical resection, clearance of all tumor from mediastinal lymph nodes is important for a favorable outcome after subsequent surgery. Repeat invasive nodal sampling by mediastinoscopy is difficult owing to the extensive mediastinal fibrosis that results from neoadjuvant therapy, which also reduces the diagnostic yield of material obtained by endoscopic fine needle aspiration. Unfortunately, results with both CT and PET have been disappointing in use of these modalities for detection of viable residual tumor and fibrotic lymph nodes after neoadjuvant chemotherapy. Further studies are needed before PET can be used routinely for assessment of tumor response.
Detection of Recurrence after Definitive Treatment
The 2003 ASCO recommendations for the treatment of NSCLC did not see a role for routine imaging in asymptomatic patients after curative treatment of NSCLC, because no rigorous randomized, controlled trials of lung cancer follow-up were conducted to show that early detection of recurrence in asymptomatic patients would significantly prolong survival. A prospective study aggressively monitoring 192 patients postoperatively with chest radiographs every 3 months and with chest CT scans and bronchoscopy every 6 months found a 71% recurrence rate; 26% of the recurrences were in asymptomatic patients. The 3-year survival rate was 13% in all patients but 31% in patients whose recurrence was detected while they were asymptomatic. These results do not take into account the lead time bias, which is known to influence survival. Large retrospective series in which traditional morphologic imaging was used for follow-up monitoring have questioned the benefits of aggressive surveillance.
At present, no published findings assessing the effect of early detection of recurrence by PET on survival are available. Detection of recurrent disease using morphologic imaging such as CT can be hampered by the effects of treatment, both surgery and irradiation, which often leave parenchymal scars, fibrosis, pleural thickening, or effusions, any of which may simulate recurrent disease.
PET imaging has been shown to be more useful than conventional imaging for diagnosing tumor recurrence, and findings can lead to major changes in management in as many as 63% of patients with suspected relapse. Several prospective studies have shown a sensitivity of 98% to 100% and a specificity of 62% to 92% for the detection of recurrent malignancy after definitive treatment with surgery, chemotherapy, or radiotherapy. Specificity of PET for detection of malignant disease is lower than at initial staging because post-therapeutic inflammation is FDG-avid, especially in the first few months after radiation therapy. This pitfall can be overcome only by careful inspection of both PET and CT images. Diffuse FDG uptake is suggestive of the inflammation associated with radiation therapy, whereas focal uptake is more suggestive of recurrence. Waiting 3 to 6 months for the inflammatory response to subside will permit detection of the focal residual FDG uptake of recurrence. For improved detection earlier on, careful inspection of the CT images may be helpful in identifying typical inflammatory changes. Findings that suggest recurrence of disease on a post-treatment PET scan should be confirmed by biopsy, to avoid treatment errors.
Uncommon Primary Pulmonary Malignancies
Some histologic subtypes of primary lung cancer are rare but typically have radiologic features that may suggest their histologic traits, as discussed next.
Sarcomatoid Carcinoma
Carcinomas with pleomorphic, sarcomatoid, or sarcomatous elements are rare. On radiologic images, these neoplasms can manifest either as large peripheral masses or as polypoid endobronchial lesions with atelectasis or postobstructive pneumonia. Calcification and cavitation are uncommon, but necrosis and hemorrhage can manifest as areas of heterogeneous attenuation on CT scan ( Fig. 1-17 ). Hilar or mediastinal adenopathy is uncommon. Pleural effusion can result from local invasion. Metastases involve sites similar to those of lung cancer: lung, liver, bones, adrenal glands, and brain.
The typical appearance of pulmonary blastoma is a large (2.5 to 26 cm in diameter), well-marginated peripheral mass. Multiple masses, cavitation, and calcification are rare. Local invasion of the mediastinum and of the pleura occurs in 8% and 25% of cases, respectively. Metastases to hilar and mediastinal lymph nodes are present in 30% of the cases after resection. Extrathoracic metastases are common and have a distribution similar to that of lung cancer.
Neoplasms of the Tracheobronchial Glands
Neoplasms of the tracheobronchial glands only rarely manifest as a peripheral pulmonary nodule, which usually is located in the central airways. Of note, these neoplasms frequently are missed on chest radiographs, because the tracheal air column and proximal bronchi often constitute a “blind spot” for many radiologists. This limitation suggests the importance of CT imaging, which readily demonstrates the airways, in an adult patient who presents with new-onset “asthma.”
Up to 80% of adenoid cystic carcinomas are confined to the trachea or main bronchi ( Fig. 1-18 ), but 10% to 15% may manifest as a peripheral pulmonary nodule. The typical radiologic appearance is that of an endotracheal or endobronchial mass, usually lobulated or polypoid, encroaching on the airway lumen. Masses can be circumferential and may manifest as diffuse stenosis. A less common manifestation is a peripheral lung nodule or mass. Although metastatic spread from adenoid cystic carcinoma has a distribution similar to that of metastatic spread from NSCLC, such spread occurs late, because this tumor exhibits slow, progressive local growth. Patients with this cancer, therefore, usually are considered to be surgical candidates, and CT is used for surgical planning. Although CT readily demonstrates the extratracheal extent of these tumors, it underestimates the longitudinal extent of the tumor. This limitation is related in part to technical factors, which can be addressed by imaging the tumor with thin slices and reconstruction in different planes, but also to the tendency of the tumor to infiltrate beneath the mucosa, which is not identifiable by CT.
