Positron Emission Tomography in Chest Diseases
Perry Gerard
Richard L. Wahl
Positron emission tomography (PET) is assuming an increasingly important role in the management of thoracic neoplasms. This technology, long only a research technique, firmly moved into the mainstream of medical imaging in the last decade of the twentieth century and the first years of the twenty-first and has gained an important role in a broad range of cancer applications, as noted in the publication of Cohade and the author.12 PET differs from anatomic imaging techniques in that it predominantly images physiology. The physiology imaged depends on the choice of the radiopharmaceutical injected, but in clinical practice, as pointed out by Brown and colleagues,6 PET currently focuses on the use of a radiolabeled glucose analog, [18F]fluorodeoxy-D-glucose (FDG), which targets the increased glucose metabolism present in most cancers of the thorax. The anatomic resolution of PET is less satisfactory than the resolution of computed tomography (CT) or magnetic resonance imaging (MRI). However, the anatomic limitations of PET imaging are being addressed by the combination of PET scanners with CT scanners into PET/CT scanners, which are undergoing greater application, as reported by Cohade and the author.12 Software fusion also allows the combination of PET images with CT or MR images to improve imaging specificity, according to Skalski and associates,72 but is less commonly applied at present than hardware-fused images using combined PET/CT scanners.
The limitations of conventional imaging for detecting and characterizing thoracic neoplasms include the following: (a) anatomic images display masses but do not generally accurately characterize what is inside the mass except to reveal whether it is water density, calcium, fat, or air and whether the blood vessels are leaky to contrast (i.e., determines contrast enhancement); (b) anatomic methods are often insensitive to small foci of tumor metastases such as in lymph nodes, although size may be a modest predictor of the presence or absence of tumor in lymph nodes; (c) anatomic methods are incapable of predicting when tumors will be responsive to a given treatment, and anatomic images are slow to change in the face of effective therapy; and finally (d) anatomic imaging methods, as I82 have stressed, may be difficult to interpret in the postoperative or post–radiation therapy setting, when it is important to separate scar from viable tumor. PET in clinical practice is generally based on the imaging of metabolic alterations located within cancers and can address most or all of these issues.
Mechanism of FDG Uptake
The most common alteration of malignant cells currently targeted in clinical practice in PET imaging is their increased use of glucose. This is traced by the radiotracer FDG. This particular compound is transported into cancer cells by glucose transporters (most notably Glut 1, which is overexpressed in cancers), phosphorylated by hexokinase within cancer cells, and then trapped within cancer cells as FDG-6-phosphate. Brown and coinvestigators6 have shown that cancer cells typically have high levels of glycolytic activity, in part owing to the increased presence of glucose transporter molecules on the cell surface and their increased levels of hexokinase compared with normal tissue. However, glucose use is not a completely specific phenomenon, and a variety of benign processes (e.g., tuberculosis, sarcoidosis, infectious processes) may have varying levels of elevated glucose use. Thus FDG PET scanning, although an excellent targeting mechanism for cancers, is not a completely specific method, as I,82 among others, have observed.
A discussion of the physics, instrumentation, and techniques of PET imaging is beyond the scope of this chapter; however, full textbooks exist on this subject, such as the one published by myself and Buchanan.83 Nonetheless, a few salient points may be briefly noted. PET imaging is somewhat similar in concept to most nuclear medicine techniques, such as bone scanning. In nuclear imaging, a radiotracer that will hone to a specific tissue or organ, based on its physical properties, is injected intravenously. As an example, in the commonly used bone scan, remodeling bone accumulates a technetium-99m (99mTc) medronate methylene diphosphonate (MDP) analog, which can then be imaged using a special camera found in nuclear medicine departments called a gamma camera; it detects the gamma rays emitted from the patient by the 99mTc. The principle for PET is similar in that a radioactive tracer that preferentially accumulates in tumor tissue owing to altered tumor metabolism is used. Rather than a gamma ray–emitting isotope, the radiotracer used is a positron emitter. The positron is a positively charged electron, which is essentially “antimatter.” The positron travels a short distance in tissue and then encounters an electron. The two combine and then annihilate, giving off two 511-KeV photons, which travel in opposite directions. These photons can then be detected by the PET scanner. The PET
scanner and its computers can determine the line on which the disintegration occurred, and then, with multiple decays and measurements, determine the point of origin of the decays using reconstruction methods like those used to generate CT scans. The term PET imaging is actually somewhat of a misnomer, because the process does not really image the positrons with a PET scanner; rather, the images obtained are of 511-KeV photons from annihilation of the positron. Several important points need to be considered: PET imaging is currently able to detect many lesions smaller than those detectable by CT because small lesions with high metabolic activity, and thus high radioactivity concentrations, can be seen with a highly sensitive PET scanner but appear normal on CT. Also, PET in its current form is not equivalent to a microscopic examination and typically does not generally detect lesions <3 mm. Indeed, many lesions that are 3 to 7 or 8 mm in diameter escape detection. Lesions are detectable with PET not only owing to lesion size but also because of lesion location, body size, tumor metabolic activity, scanner performance, and the duration of the scan acquisition. Patient motion can also affect detection by blurring the image signal. Nonetheless, PET can detect lesions that are typically not detected using anatomic methods, hence its higher sensitivity than that of anatomic imaging in a wide array of neoplasms.
scanner and its computers can determine the line on which the disintegration occurred, and then, with multiple decays and measurements, determine the point of origin of the decays using reconstruction methods like those used to generate CT scans. The term PET imaging is actually somewhat of a misnomer, because the process does not really image the positrons with a PET scanner; rather, the images obtained are of 511-KeV photons from annihilation of the positron. Several important points need to be considered: PET imaging is currently able to detect many lesions smaller than those detectable by CT because small lesions with high metabolic activity, and thus high radioactivity concentrations, can be seen with a highly sensitive PET scanner but appear normal on CT. Also, PET in its current form is not equivalent to a microscopic examination and typically does not generally detect lesions <3 mm. Indeed, many lesions that are 3 to 7 or 8 mm in diameter escape detection. Lesions are detectable with PET not only owing to lesion size but also because of lesion location, body size, tumor metabolic activity, scanner performance, and the duration of the scan acquisition. Patient motion can also affect detection by blurring the image signal. Nonetheless, PET can detect lesions that are typically not detected using anatomic methods, hence its higher sensitivity than that of anatomic imaging in a wide array of neoplasms.
Patient Preparation for FDG PET Scanning
Several points regarding patient preparation for FDG PET scanning should be considered. In general, for tumor imaging, patients are evaluated in the fasting state, typically after fasting ≥4 hours. This is designed to lower serum insulin and glucose levels and to minimize cardiac uptake of FDG. Patients who have recently eaten can have altered distribution of FDG, and their PET scans may show increased targeting to muscle and other tissues, which can be confusing. Patients with severely altered glucose metabolism may not be ideal candidates for FDG PET scanning, but those with reasonably well controlled diabetes can be imaged with FDG PET. Increased serum glucose levels can reduce tumor targeting and may, in some cases, make tumor detection extremely difficult. Lung tumors typically have high enough FDG uptake that they will be detected even if glucose levels are elevated, but many institutions will not perform PET scans if the fasting glucose level is >200 mg/dL. Patients with such glucose levels may need insulin therapy to optimize their tumor visualization, but care must be taken because FDG administration just after insulin has been given may be less accurate than PET performed in the fasting state. This area is under study.
Conduct of PET Scanning
Typically, patients who have FDG PET scans are injected with the radiotracer and then undergo imaging 45 minutes to 1 hour after injection. Scans can take from 20 to 60 minutes to acquire. Technologies continue to improve, and scanning durations are shortening. However, a PET scan takes significantly longer than a CT scan with a multidetector unit, and the patient must lie relatively still in the scanner for moderate periods of time. Patients breathe freely during the PET study, and respiratory motion can degrade visualization of small lesions in the lung and upper abdomen, where respiration effects may alter lesion location during the scan. Attempts to gate for the respiratory cycle are under study by Nehmeh and colleagues,59 among others, but are not routine at this time. Such methods potentially may lead to detectability of smaller lesions than ungated PET images but may lengthen the acquisition times.
Evaluation of the PET Scan
PET is an inherently quantitative technique; this ability can be used to determine the absolute uptake of FDG or other tracers in the body. The standardized uptake value (SUV) is a parameter that is sometimes used to describe lesion uptake. In general, the higher the SUV, the more likely it is that a lesion will prove to be malignant. Malignant lesions of the thorax in general have very high glucose metabolism levels. SUV is discussed in more detail under “Solitary Pulmonary Nodules,” later in this chapter.
For PET scanning to be an accurate quantitative parameter, measurement of body thickness and thus photon attenuation must be performed. Body thickness is usually measured by a so-called transmission scan, which is a low- or high-quality CT scan of the chest and abdomen. The length of time to acquire this scan can vary. With modern PET/CT scanners, the transmission imaging may take only 20 to 30 seconds. With older PET scanners, transmission measurements can take 15 to 30 minutes and are performed with a high-energy source providing only limited anatomic resolution. Obviously shortening the transmission scan is an important tool to shortening the duration of the study. Shorter study durations result in more cost-effective studies in general because more scans can be done in a given period of time.
PET scanners are typically relatively expensive devices and cost between $1 million and $2.5 million each. Instruments at the upper end of the price range may well include an integrated CT scanner, which allows more precise registration of the PET images to anatomic structures. Use of PET/CT scanners is growing rapidly, and for most manufacturers, PET/CT scanners are now the dominant proportion of their PET scanner sales. This technology is in rapid evolution, but the ability to identify tumors using the sensitivity of PET and to locate them precisely using CT is a potent technology that is gaining rapid acceptance in many major medical centers.
Acquisition of an Appropriate PET Scan
PET scans can be performed in a variety of ways, but the most common approach for thoracic neoplasms is for the scan to extend from the midneck through the proximal thighs. In this way, the supraclavicular and cervical lymph nodes and the entire thorax, liver, adrenals, abdomen, and proximal femoral bones are evaluated. A rather thorough assessment of the entire body and not just of the thorax is achieved. Patterns of imaging may vary in different practices, but the so-called whole-body PET study is commonly performed as just described. It should be realized that the whole-body study often does not include the brain and often not the entirety of the lower extremities. This differs from one institution to another, but because of time constraints, the lower extremities are often excluded from the field of view. It is important to consider this factor in contemplating the use of PET to replace bone scanning, a study in which the entire skeleton is adequately imaged. The uses of PET in lung and esophageal cancers are discussed separately.
Use of FDG Pet in Thoracic Neoplasms
Lung Cancer
Solitary Pulmonary Nodules
Solitary pulmonary nodules (SPNs) are reasonably common and typically represent pulmonary parenchymal lesions between 1 and 3 cm in diameter. These lesions are typically not calcified and their age is indeterminate. Although some level of characterization of risk for neoplasm can be achieved by measuring lesion size, patient age, smoking history, and characteristic of the lesions and their density on CT, it is in fact quite difficult to assign a definitive and precise diagnosis in an individual patient. Thus these SPNs carry with them an intermediate likelihood of containing cancer, typically ranging from 20% to 80%.
In the past, SPNs represented about one-third of newly diagnosed lung cancers, but with the greater use of screening programs, SPNs in some centers now represent a larger fraction of the patient population with newly diagnosed lung cancer. Accurate diagnosis and intervention in these patients, according to Chang and Sugarbaker,9 is quite important because these tumors represent the most curable lung cancers. Historically, assessment of SPNs has relied on CT with sequential CT scans, fiberoptic bronchoscopy, needle aspiration biopsy, or excision. Needle aspiration biopsy, although popular in some institutions, has a sampling error problem that may result in false-negative exams, and there is a risk for pneumothorax. This technology is being performed less frequently as noninvasive techniques gain favor.
A relatively large number of reports have evaluated the utility of PET imaging with FDG of SPNs and have been summarized in several analyses, such as the meta-analysis published by Gould and associates.26 These assessments are based on the expectation and observation that most lung cancers within SPNs have higher levels of glycolytic metabolic activity than benign pulmonary nodules. A number of reports dating from the early 1990s have shown sensitivity for cancers within pulmonary nodules to be in the range of 90% to 100%, with specificities typically in the range of 60% to high 80%. The meta-analysis by Gould and associates26 of 450 pulmonary nodules reported the mean joint operating sensitivity and specificity for PET to be about 91.2%. Using more typical readings to optimize sensitivity, a sensitivity of 96.8% for cancers could be achieved with a specificity of about 77.8%. As noted by Gould and associates,26 most interpreters of PET studies err on the side of higher sensitivity so as to try to achieve at least 95% sensitivity in lesion detection at the expense of slightly lower specificity. There are some data to suggest that the utility of FDG PET will differ depending on the prevalence of inflammatory and granulomatous disease in the population. As an example, in a patient population with a high frequency of histoplasmosis, the positive predictive value of PET was lowered substantially owing to the common occurrence of false-positive scans of nonmalignant, inflammatory pulmonary lesions, as discussed by Croft and colleagues.14
In general, SPNs are assessed by PET imaging about 1 hour after FDG injection by visual assessment to determine whether there is increased FDG uptake in the lesion that is greater than the activity in the cardiac blood pool. This is usually done on attenuation-corrected images. It is recognized that very small nodules (<1 cm) may have somewhat less FDG uptake than blood pool activity and yet be malignant. Thus many investigators tend to report lesions <1 cm as positive even if uptake is slightly less than that of the blood pool. The PET techniques are best for characterizing lesions >1 cm. The data in lesions 7 mm to 1 cm in size are relatively scanty. Clearly, in lesions <7 mm, PET must be used with caution because the negative PET scan is unlikely to have the same high sensitivity and negative predictive value that it does in larger lesions. This is because of the inherent detection limitations of small lesions with current PET technology. CT of the thorax remains necessary for pulmonary nodules <7 mm. It is also clear that lesions near the lung bases may have somewhat higher background activity around them than lesions in the upper lobes and that the lower-lobe lesions may have more motion, which can blur the images. Corrections of quantitative SUVs by lesion size are also feasible; Hickeson and coinvestigators31 suggest that this may improve the accuracy of PET in small nodules.
How PET with FDG is used in evaluating pulmonary nodules may vary somewhat among various centers. It is clear, however, that if the pretest likelihood of disease is relatively low, the posttest probability of cancer after a negative PET scan can be very low, in the range of 1% to 2%. In such patients, only very limited follow-up imaging is necessary if observation is chosen. However, a conservative approach would suggest that some follow-up imaging, possibly anatomic, 6 to 12 months after imaging might be appropriate even when there is a negative PET scan, so as to exclude the growth of an FDG-negative tumor. In patients with larger primary lesions, extensive smoking history, and other risk factors, a negative scan may not be sufficiently predictive of the absence of disease to exclude tumor completely. Follow-up imaging may be required with anatomic methods to determine whether a false-negative result has occurred. False-positive results can occur in patients with tuberculosis, granulomatosis, and sometimes other inflammatory processes. False-positive results are part of the known limitations of FDG PET of SPNs, as emphasized by Hickeson and colleagues.31
Our group, like others, has found that there is higher FDG background activity in the lung bases and in the posterior portions of the lungs. Small lesions near the lung bases may be somewhat more difficult to detect than larger lesions. Most institutions, although capable of performing SUVs, do not report the SUV but rather interpret the images qualitatively—that is, whether the lesion has increased glucose metabolism relative to surrounding background tissues. Determinations of SUVs are quite useful, however, for evaluating lesions after therapy and are so used in some centers.
Quantitative SUVs have been used in PET imaging to try to separate malignant and benign pulmonary lesions. In some studies, SUVs >2.5 are associated with a higher probability of cancer being present than lesions with an SUV <2.5. But the SUV is affected by a variety of other factors, including time from injection, quality of injected dose, patient body mass, and region-of-interest selections. Pulmonary nodules of this size, however, may have considerably varying levels of uptake. Because the variability of SUV determinations is typically about 10%, lesions with an SUV of 2.3 to 2.7 may in fact have very similar true SUVs, but the difference is largely based on statistical variation. For this reason, it is probable that absolute reliance on SUV will be a mistake in lesion assessments. Thus most experienced laboratories use mainly qualitative analysis to assess pulmonary nodules. When
quantitative SUV is applied, there is evidence, presented by Hickeson and associates,31 that higher accuracy may be achieved by incorporating corrections for partial volume effects based on lesion size on CT.
quantitative SUV is applied, there is evidence, presented by Hickeson and associates,31 that higher accuracy may be achieved by incorporating corrections for partial volume effects based on lesion size on CT.
It is important to compare lesion size in PET images with CT. This should be done with combined PET/CT scanners or with images obtained separately with the two modalities. Pulmonary nodule size is extremely important for accurate interpretation of PET images of the lungs for the presence or absence of lung cancer. There have been some limited direct comparative studies evaluating the accuracy of PET versus aspiration biopsy in SPNs. In these studies, aspiration biopsy was less sensitive than PET for detecting cancer in pulmonary nodules. This lower sensitivity was a result of inadequate sampling by the aspiration biopsy with quite sufficient noninvasive sampling achieved by PET imaging. The effectiveness of FDG PET in SPNs has been demonstrated but is not likely to be cost-effective as a method of avoiding a thoracotomy if there is a high prevalence of cancer within the nodules. In general, avoidance of thoracotomy or other major invasive procedures is the mechanism by which PET saves money in patient management. If PET only adds to the number of required procedures and is not a replacement for other procedures, it can add to patient care costs. Nonetheless, there is increasing application of PET in the evaluation of SPNs because of its overall accuracy, in the 90% range.
A limitation of FDG PET in SPNs is its inability to detect a small fraction of primary lung cancers. The lung cancers least likely to be detected by FDG PET, as reported by Marom and associates,53 include bronchioloalveolar carcinoma (tumors that typically have low cellularity and low glycolytic activity), some neuroendocrine tumors such as carcinoids, and mucinous tumors in which much of the tumor bulk is made of mucin and not viable tumor cells. Other false-negative reports include very small tumors. False-positive reports can occur most commonly in the inflammatory conditions, with tuberculosis being among the most problematic, especially in areas where tuberculosis is endemic. Some data published by Matthies and coinvestigators54 indicate that SPNs that have an increase in FDG uptake between 1 and 3 hours after tracer injection are malignant, whereas those with a decline are more likely benign. Recently the added value of PET to screening CT programs was demonstrated by Lowe,45 suggesting that the value of PET in assessing lung nodules found by standard methods might also be applicable for lung nodules detected by screening programs. Although most publications have focused on primary lung cancers, Dalrymple-Hay and collaborators15 suggest that PET appears to be equally good in detecting metastatic cancers to the lung such as melanoma, albeit with size limitations on its overall performance.
Local and Regional Nodal Staging in Lung Cancer
After lung cancer is diagnosed, whether tumor involves the regional lymph nodes or does not is an important consideration for prognosis and choice of management. CT has been extensively used in the evaluation of locoregional lymph nodes for their involvement with tumor for some time, and, although a useful technique, CT is not particularly sensitive or specific in determining the presence or absence of metastases to mediastinal lymph nodes. The 1-cm criterion used as a threshold value to separate benign and malignant disease is only a fair predictor of whether mediastinal tumor involvement is present and MRI has not, to date, been more accurate, according to Dwamena18 and Gould27 and their colleagues. This only modest accuracy of CT is related to the limited ability of size alone to determine whether tumor is present or absent in a given structure. Some nodes are enlarged as a result of benign processes, whereas others are of normal size but still involved with tumor.
In 1994, I and my associates84 reported PET to be significantly more accurate than CT in mediastinal nodal staging in a prospective trial. We compared PET directly to CT to make this conclusion. We also used computer fusion software to combine digitally the CT and PET and showed that this did not significantly increase the accuracy of the PET imaging using existing software. A substantial number of reports have come forth in the past decade indicating PET to be more accurate than CT for mediastinal nodal staging. A meta-analysis by Dwamena and coinvestigators18 of 14 studies with a total of 514 PET patients and of 29 CT studies with a total of 2,226 patients revealed a higher sensitivity of 91% for PET versus 79% for CT and a higher specificity of 77% for PET versus 60% for CT in the assessment of mediastinal lymph nodes. Furthermore, the overall accuracy of PET was superior to that of CT in this analysis; the observed characteristics were clearly and significantly superior for PET over CT scans. It must be realized, however, that these data, although impressive, do not indicate PET to be equivalent to histologic sampling. A more recent meta-analysis by Gould and colleagues27 that included 39 studies also showed PET to be more accurate than CT for identifying mediastinal nodal tumor involvement (p <0.001). For CT, median sensitivity and specificity were 61% (interquartile range, 50%–71%) and 79% (interquartile range, 66%–89%), respectively. For FDG PET, median sensitivity and specificity were 85% (interquartile range, 67%–91%) and 90% (interquartile range, 82%–96%), respectively. Fourteen studies provided information about the conditional test performance of CT and FDG PET. FDG PET was more sensitive but less specific when CT showed enlarged lymph nodes (median sensitivity, 100% [interquartile range, 90%–100%]; median specificity, 78% [interquartile range, 68%–100%]) than when CT showed no lymph node enlargement (median sensitivity, 82% [interquartile range, 65%–100%]; median specificity, 93% [interquartile range, 92% to 100%]; p = 0.002). Optimal methods to integrate nodal size with PET findings and risk for cancer continue to evolve.
A review of the literature by Toloza and associates77 provided similar results and showed PET to be more accurate than CT or transesophageal endoscopic ultrasound (EUS) studies. Pooled sensitivities and specificities for staging the mediastinum were as follows: (a) for CT scanning: sensitivity, 0.57 (95% confidence interval [CI], 0.49–0.66) and specificity, 0.82 (95% CI, 0.77–0.86); (b) for PET scanning: sensitivity, 0.84 (95% CI, 0.78–0.89) and specificity, 0.89 (95% CI, 0.83–0.93); and (c) for EUS: sensitivity, 0.78 (95% CI, 0.61–0.89) and specificity, 0.71 (95% CI, 0.56–0.82). The EUS technique is still in evolution, but it may be very useful when combined with PET.
Although sensitivities of 85% to 90% are encouraging, this means that about 10% to 15% of nodal metastases from lung cancer in the mediastinum will not be detected by PET. Further, the specificity of 80% to 90% for PET indicates that a moderate number of false-positive results occur. My impression and that of many in my field is that the specificity of PET can differ based on the prevalence of inflammatory disease in the patient population. In patient populations with extensive tuberculosis,
PET may be less specific than in developed countries, where infectious diseases are well controlled. Inflammatory diseases such as histoplasmosis and sarcoidosis can have false-positive results as well, causing the performance of PET to vary based on geographic locale and endogenous prevalence of granulomatous disease. Nonetheless the sensitivity of PET is high and the specificity is reasonably high, but a negative PET does not categorically exclude nodal metastases in the mediastinum and a positive PET does not categorically prove that tumor is present.
PET may be less specific than in developed countries, where infectious diseases are well controlled. Inflammatory diseases such as histoplasmosis and sarcoidosis can have false-positive results as well, causing the performance of PET to vary based on geographic locale and endogenous prevalence of granulomatous disease. Nonetheless the sensitivity of PET is high and the specificity is reasonably high, but a negative PET does not categorically exclude nodal metastases in the mediastinum and a positive PET does not categorically prove that tumor is present.
There is some limited recent evidence to suggest that the quantitative assessment of FDG uptake levels in mediastinal lymph nodes has some predictive value for the presence or absence of mediastinal metastases. In a study in Korea where there is a relatively high frequency of granulomatous disease, SUVs >3.4 were found to be a good separator between malignant and benign mediastinal disease, with an accuracy of about 85% (greater than that of CT), according to Kang and coworkers.37 My practice is to recommend that the most advanced site of apparent metastatic disease identified by PET be confirmed histologically to avoid erroneously placing a patient into a nonsurgical group who in fact would otherwise be potentially curable with surgical intervention.
PET/CT technology is reasonably new but has been shown to be more accurate than PET alone or PET visually compared (but not fused) to a concurrent CT in staging the mediastinum and in assessing T stage. In the study by Lardinois and associates,41 integrated PET/CT provided additional information in 20 of 49 patients (41%) beyond that provided by conventional visual correlation of PET and CT. Integrated PET/CT had better diagnostic accuracy than the other imaging methods. Tumor staging was significantly more accurate with integrated PET/CT than with CT alone (p = 0.001), PET alone (p<0.001), or visual correlation of PET and CT (p = 0.013); lymph node staging was also significantly more accurate with integrated PET/CT than with PET alone (p = 0.013). In metastasis staging, integrated PET/CT increased the diagnostic certainty in 2 of 8 patients reported by Lardinois and colleagues.41
Similar results were reported in a study by Antoch and associates3 of 27 patients with non-small-cell lung cancer (NSCLC) using PET/CT in which overall tumor stage was correctly classified as 0 to IV with CT in 19 patients, with PET in 20 patients, and with PET/CT in 26 patients. Differences in the accuracy of overall tumor staging between PET/CT and CT (p = 0.008) and between PET/CT and PET (p = 0.031) were significant. Primary tumor stage was correctly determined in more patients with PET/CT than with either PET alone or CT alone. Accuracy of PET/CT for regional nodal staging was 93%, for PET 89%, and for CT 63%. Clearly PET is a very accurate method in lung cancer assessment, but it appears that the fused PET/CT study adds incremental value and some modest improvement in accuracy.
Thus, in many institutions, PET or PET/CT is being performed at the initial assessment of lung cancers to determine whether the patient should have surgery. In many centers, patients with a negative CT scan of the mediastinum and a negative PET scan will proceed directly to surgery and will not have a mediastinoscopy performed. In general, if the PET scan is positive, regardless of the CT, tissue sampling is in order. Obviously, positive PET and CT scans in the mediastinum are considered very suspicious for metastatic disease; however, confirmation by tissue is again recommended by Silvestri and coworkers71 in such cases to avoid the infrequent situation when increased tracer uptake in the mediastinum represents a false-positive finding. In general, a false-negative PET in the mediastinum is associated with low tumor burden in lymph nodes, and it is probable though not proven that such findings are associated with a better prognosis than more bulky metastatic involvement.
Staging Distant Metastatic Disease
Lung cancer with mediastinal nodal metastases is only infrequently cured. Lung cancer with distant metastases is not surgically curable (with rare exceptions; see Chapter 110), and if distant metastases are identified when the patient presents, surgery for “cure” will typically be canceled.
The frequency of distant metastases at disease presentation is reasonably common at 10% to 20%, and PET has assumed an increasingly important role in staging for metastatic lung cancer outside of the thorax. Several studies, such as those of Lowe45 and Mac Manus and Hicks,49 have shown that PET can detect distant metastases following normal standard staging in 5% to 20% of patients, depending on the stage of the primary tumor and intensity of the conventional staging evaluation. Further, Stroobants and colleagues74 have noted that PET can offer confidence that equivocal lesions seen on standard staging methods are indeed benign in a significant fraction of cases.
Areas of particular strength for PET beyond lymph nodal assessments include (a) the evaluation of the adrenal gland, where the accuracy of PET has been reported by Gupta and associates28 to be about 90%, and (b) assessing the presence of metastatic disease in the bones, where the predominantly osteolytic lesions of lung cancer are well visualized on PET and false-positive findings appear to occur with lower frequency than on bone scans (such as in degenerative change that are often positive on bone scan but negative on PET). The bone scan remains a useful test, however. A recent comparative study of PET with FDG and bone scan revealed a sensitivity, specificity, and positive and negative predictive value of 81%, 78%, 34%, and 93%, respectively, for bone scan and for FDG PET (73% [p = 0.81], 88% [p = 0.03], 46% [p = 0.5], and 97% [p = 0.04], respectively). Gayed and coworkers23 have reported that FDG PET scans demonstrate significantly higher specificity and negative predictive values than bone scans in evaluating bony metastases from lung cancer. Nonetheless, because some bone metastases are not detected by FDG PET alone, it is probably not appropriate to eliminate the bone scan from the lung cancer staging workup until more data are available.
PET with FDG has also shown promise in characterizing pleural effusions as malignant or benign, based on the intensity of FDG uptake. Gupta and associates29 have suggested that those effusions with intense FDG uptake are much more likely to be malignant than those with modest or faint FDG uptake. Another area in which PET with FDG is particularly helpful is in the assessment of liver metastases. PET with FDG appears to be less likely to have false-positive results in the liver than CT. A study from Duke University, reported by Marom and colleagues,52 in which both PET and CT of 100 lung cancer patients were performed showed that PET detected half as many liver lesions as CT, but all the hepatic lesions detected by PET were malignant, whereas half of the lesions detected by CT were not malignant. Thus, PET had fewer false-positive results. Meta-analysis of PET in lung cancer for evaluation of liver lesions has indicated PET to be more
sensitive than CT in the evaluation of metastases. Nonetheless, CT of the liver remains a useful technique; some lesions will not be detected on PET, although they will be infrequent, at least based on the experience of Kinkel and coinvestigators38 in a broad array of cancers metastatic to the liver.
sensitive than CT in the evaluation of metastases. Nonetheless, CT of the liver remains a useful technique; some lesions will not be detected on PET, although they will be infrequent, at least based on the experience of Kinkel and coinvestigators38 in a broad array of cancers metastatic to the liver.
As indicated earlier in this chapter, brain imaging is often not performed as part of the PET scan, and most whole-body protocols exclude the brain from the imaging field. This is because there is intense FDG uptake within the normal brain that can overlap with the intensity of FDG uptake in brain tumors. Thus a significant fraction of brain metastases from lung cancer will not be detected by PET imaging owing to the intense metabolic activity in normal brain. If brain metastases are strongly suspected, it is important to perform anatomic imaging with contrast enhancement to rule out metastases. In a study from Vanderbilt University, screening for cerebral lesions in patients with body malignancy had little clinical impact. Unsuspected cerebral or skull metastases were detected in only 0.4% of patients (4 of 1,026) in a study reported by Ludwig and colleagues.46 Thus, most PET protocols for brain imaging do not include imaging of the brain.
Better staging accuracy should result in better decisions in individual patient management. Few studies have been done to prove this with any particular imaging method. However, a randomized study has been published by van Tinteren and associates79 comparing the frequency of futile thoracotomies in apparent newly diagnosed lung cancer patients who underwent PET scanning with those who did not undergo PET for staging. Ninety-six patients were randomly assigned to a conventional workup lacking PET and 92 to a conventional workup including PET. Two patients in the conventional workup + PET group did not undergo PET. Eighteen patients in the conventional workup group and 32 in the conventional workup + PET group did not have thoracotomy. In the conventional workup group, 39 (41%) patients had a futile thoracotomy, compared with 19 (21%) in the conventional workup + PET group (relative reduction, 51%; 95% CI, 32%–80%; p = 0.003). In this study, the addition of PET to the conventional workup prevented unnecessary surgery in 1 of 5 patients with suspected NSCLC. This reduction in the number of futile thoracotomies would be expected to be cost-effective in most health-care systems. Elimination of surgery in patients with unsuspected distant metastases is appropriate, and for this reason, a very large fraction of primary lung cancer patients in many centers receive PET scans as a part of the presurgical workup.
Most NSCLCs, of both squamous and adenocarcinoma histology, have similar levels of FDG uptake, although it must be cautioned, as noted by Marom and colleagues,53 that bronchioloalveolar carcinoma generally has lower FDG uptake than the other types of NSCLCs, as do carcinoid tumors. Untreated small-cell lung carcinomas (SCLCs) frequently have very intense FDG uptake, often associated with a large mass lesion. Although the data for SCLC are less comprehensive than for NSCLC, FDG uptake in SCLC, as reported by Zhao and associates,90 is sufficient to allow visualization of active disease in most tissues, including pleural recurrences and in the bone marrow. For this reason, SCLC imaging with FDG increasingly is being applied. Because some of the decisions in patients with SCLC usually involve determining whether radiation therapy is required, knowing the precise extent of disease and whether it would extend beyond the radiation ports is important. My group is seeing increased use of PET for this clinical indication. Data available to date indicate that virtually all untreated SCLC can be imaged with FDG PET and that the intensity of FDG uptake, according to Pandit and coinvestigators,62 is associated with prognosis, with those lesions having the highest uptake having a less poor prognosis than those not intensely FDG-avid. The role of PET in outcome and management of SCLC is in evolution.
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
