Introduction
Positron emission tomography (PET) with 18 F-fluorodeoxyglucose (FDG) is a noninvasive imaging technique with several important applications in respiratory medicine. PET has long been used to evaluate inflammatory conditions such as sarcoidosis or idiopathic pulmonary fibrosis; for example, in small series of about 20 patients, the extent and/or activity of these diseases could be assessed more precisely by PET/ computed tomography (CT) than by 67 Gallium single-photon emission computed tomography scintigraphy. PET has also been studied in patients with posttransplantation lymphoproliferative disease, in which a possible role in staging and follow-up was suggested.
The vast majority of data and clinical applications of PET, however, pertain to patients with respiratory malignancies, such as lung cancer or mesothelioma, and these are the main subjects of discussion in this chapter. Because FDG is by far the most commonly used tracer for this purpose, the term PET refers to FDG-PET unless stated otherwise.
Principles
PET Camera
A PET camera produces three-dimensional images that represent the distribution of radioactivity within the body of a patient. Any molecule labeled with a positron-emitting radioisotope can be used to generate PET images. The PET camera consists of a full ring of several thousand scintillation detectors to generate the image, resulting in higher sensitivity to radioactivity and better spatial resolution than the conventional gamma camera. The spatial resolution of contemporary PET cameras is around 4 mm, allowing accurate characterization of lesions larger than 8 mm.
Modern PET cameras are hybrid systems in which a PET camera is combined with either a CT or magnetic resonance camera. Hybrid PET/CT cameras are now considered standard, whereas PET/magnetic resonance is an emerging technology. Compared to PET alone, hybrid PET/CT cameras provide three main advantages: (1) CT-based attenuation correction (AC) is applied to the PET image to correct for absorption by the body of the patient; (2) there is increased accuracy of the exact position of the lesion and morphologic characterization of the underlying correlate, reducing equivocal findings; and (3) the combined image increases the confidence of the reporting physician.
Typical scan times for modern PET/CT are in the order of 15 minutes for a skull-to-thigh, “whole body,” image. Unless a separate dedicated contrast-enhanced CT scan is already available, PET/CT is preferably combined with high-dose, contrast-enhanced diagnostic CT than with low-dose CT; with high-dose CT, the PET/CT is more precise for tumor-node-metastasis (TNM) staging because of better AC and localization. It has been demonstrated that the use of oral or intravenous contrast agents does not induce clinically significant changes in the PET images.
Metabolic Tracer: FDG
For clinical cancer imaging the glucose analogue FDG is by far the most common tracer. Its use is based on the increased cellular uptake of glucose, due to both an increased expression of glucose transporter proteins and a much higher rate of glycolysis of cancer cells. FDG, a glucose analogue in which the oxygen molecule in position 2 is replaced by a positron-emitting fluorine-18 atom, undergoes the same uptake as glucose but is metabolically trapped and sequestered in neoplastic cells after phosphorylation by hexokinase. The radiation dose for a typical examination is in the order of 5 to 8 mSv, comparable to the effective dose of a diagnostic chest CT (7 to 7.5 mSv). The FDG uptake is generally expressed as the standardized uptake value (SUV), a semiquantitative measure of FDG-uptake that expresses the uptake of a lesion as a function of the total injected dose.
Interpretation of PET Images
For diagnosis and staging, visual analysis relies on the detection of foci with activity higher than background not caused by physiologic processes, both for the discrimination of nodules and for the evaluation of mediastinal involvement. Non-AC images should be examined to detect small lung lesions, because non-AC images have better contrast for such nodules than AC images. High physiologic FDG uptake is present in the brain, kidney, and urinary tract (urinary excretion) and can be present in the heart. The high uptake in the brain interferes with lesion detection. There is a low degree of physiologic uptake of FDG in the other intrathoracic structures.
False-positive findings are possible, because FDG uptake is not tumor specific and can be found in all active tissues with high glucose metabolism, particularly in sites of inflammation. Therefore FDG-positive findings, especially if isolated and decisive for patient management, require confirmation. The differentiation between metastasis, a benign or inflammatory lesion, and even an unrelated second malignancy should be established by other tests or tissue biopsy. The major causes of false-positive results in chest pathologic conditions are infectious, inflammatory, and granulomatous disorders ( Table 21-1 ) and recent medical procedures.
| FALSE-NEGATIVE FINDINGS |
| Lesion dependent |
| Small tumors (<8-10 mm) |
| Ground-glass opacity neoplasms (adenocarcinoma with lepidic pattern) |
| Carcinoid tumors |
| Technique dependent |
| Hyperglycemia |
| Paravenous FDG injection |
| Excessive time between injection and scanning |
| FALSE-POSITIVE FINDINGS |
| Infectious-inflammatory lesions |
| (Postobstructive) pneumonia—abscess |
| Mycobacterial or fungal infection |
| Granulomatous disorders (sarcoidosis, granulomatosis with polyangiitis [Wegener granulomatosis]) |
| Chronic nonspecific lymphadenitis |
| (Rheumatoid) arthritis |
| Occupational exposure (anthracosilicosis) |
| Bronchiectasis |
| Organizing pneumonia |
| Reflux esophagitis |
| Iatrogenic causes |
| FDG embolus |
| Invasive procedure (puncture, biopsy) |
| Talc pleurodesis |
| Radiation esophagitis and pneumonitis |
| Bone marrow expansion after chemotherapy |
| Colony-stimulating factors |
| Thymic hyperplasia after chemotherapy |
| Benign mass lesions |
| Salivary gland adenoma (Warthin) |
| Thyroid adenoma |
| Adrenal adenoma |
| Colorectal dysplastic polyps |
| Focal physiologic FDG uptake |
| Gastrointestinal tract |
| Muscle activity |
| Brown fat |
| Unilateral vocal cord activity |
| Atherosclerotic plaques |
False-negative findings are less common and may be due to lesion-dependent or technical factors (see Table 21-1 ). A critical mass of metabolically active malignant cells is required for PET detection. Therefore careful interpretation is warranted in tumors with low FDG uptake, such as small-sized very well differentiated adenocarcinoma, adenocarcinoma with lepidic growth, or carcinoid tumors. In addition, even in tumors with high FDG uptake, lesions less than 8 mm may prove falsely negative due to the limitations in spatial resolution; in the lower lung fields, due to the greater respiratory motion there, the detection limit may even be 10 mm. One interfering factor inherent to the technique is a high blood glucose level, which should be checked and be within an acceptable range (typically 60 to 180 mg/dL) before tracer injection.
Diagnosis
Noncalcified solitary pulmonary nodules (SPNs) are common findings on chest radiograph or CT examination in clinical practice and have become even more frequent with the recent interest in low-dose chest CT for early lung cancer detection. Initially, PET studies in the diagnosis of SPNs used a threshold maximum SUV (SUV max ) above 2.5 for the diagnosis of malignancy. Applying this criterion, overall sensitivity, specificity, and positive and negative predictive values of 96%, 78%, 91%, and 92%, respectively, were reported in a meta-analysis based on series with nodules larger than 1 cm ( Fig. 21-1 ).
However, more recently the use of an SUV max level below 2.5 to exclude malignancy has been challenged as too restrictive. It is true that solid malignant lesions of at least 1 cm will usually have an SUV max above 2.5, but smaller cancers, lesions with ground-glass appearance on CT (e.g., the lepidic type of adenocarcinoma ) ( Fig. 21-2 ), or tumors with low metabolism (e.g., carcinoid tumors ) may have an SUV max below 2.5. In a large prospective study of PET/CT scans of indeterminate lesions, SPNs smaller than 2.5 cm were found to have a 24% chance of being malignant when the SUV max was between 0 and 2.5, 80% if between 2.6 and 4.0, and 96% if 4.1 or more.
Rather than using a fixed SUV max criterion as a threshold, the visual information from PET images—lesions with any increased FDG uptake being potentially malignant—should be added to a comprehensive nodule assessment based on clinical characteristics such as smoking and age, CT imaging characteristics such as appearance (ground glass, semisolid, or solid) and margins, and growth pattern if available. Using this approach, the chance of identifying an SPN as malignant is improved compared with just using the information from PET alone. The benefit of PET in this setting was confirmed in a series of 106 radiologically indeterminate SPNs, of which 61 were malignant ; PET improved the accuracy over a prediction model that did not incorporate PET data by 13.6%.
Staging
TNM Staging
The TNM staging system classifies malignant tumors according to the extent of the primary tumor (T), the spread to locoregional lymph nodes (N), and the presence of distant metastasis (M); as a result, lung cancer patients of different TNM subsets with similar prognoses can be grouped into stages. Stage is the most important factor in prognosis and choice of treatment, which means that reliable noninvasive methods for accurate staging are extremely important. CT scans, endoscopic techniques, and surgical staging procedures are key factors, but the addition of PET to these conventional methods has been shown to improve the staging process substantially; PET greatly aids in distinguishing patients who are candidates for therapy with curative intent, such as surgical resection or intense multimodality treatment, from those who are not.
The T Factor
Modern multislice CT images allow detailed evaluation of the anatomic relationships among the tumor and the lung fissures, which may determine the type of resection, and among the tumor and both the mediastinal structures and the pleura and chest wall. In addition, integrated PET/CT images may enhance precise definition of chest wall and mediastinal infiltration or correct differentiation between tumor and peritumoral inflammation or atelectasis ( Fig. 21-3 ).
The N Factor
It has been clear since the initial studies that the addition of PET to CT results in more accurate lymph node staging than CT alone, with an overall sensitivity of 80% to 90% and a specificity of 85% to 95% for the detection of pathologic nodes. In addition, the absence of mediastinal lymph node disease on PET/CT has a high negative predictive value, so that invasive lymph node staging tests can often be omitted, allowing these patients to proceed straight to surgical resection. However, there are limitations to relying on negative PET results; one should have less confidence in negative PET results in cases of a primary tumor larger than 3 cm, insufficient FDG uptake in the primary tumor, a centrally located tumor, or concurrent hilar nodal disease that may obscure coexisting N2-disease on PET. On the other hand, positive PET/CT findings determine the location of suspect lymph nodes and thereby help to direct tissue-sampling procedures, such as endobronchial ultrasonographically-guided transbronchial needle aspiration or cervical mediastinoscopy. Because of false-positive images in lymph nodes—based on the conditions listed in Table 21-1 —proof of lymph node involvement in pathologic processes should be sought in most patients with positive mediastinal nodes on PET, except those with obvious bulky nodes on imaging.
The M Factor
PET added to CT is almost uniformly superior to CT alone, except for brain imaging, where the sensitivity for detecting lesions is unacceptably low due to the high glucose uptake of normal surrounding brain tissue. For detecting extrathoracic metastases, the pooled sensitivity and specificity for PET/CT were 77% (95% confidence interval [CI], 47% to 93%) and 95% (95% CI, 92% to 97%), respectively, in a recent meta-analysis. CT and especially magnetic resonance imaging remain the methods of choice for brain imaging.
For bone metastases, PET is more accurate than 99m Tc methylene diphosphate bone scanning: sensitivity is at least as good (90% to 95%) and specificity is far better (95% versus 60% for bone scan). Limitations are that PET only images from the head to just below the pelvis and thus will miss lesions outside this range and that PET may not detect osteoblastic lesions, which nonetheless are rare in untreated lung cancer. For adrenal gland metastases, PET has a high sensitivity, so that an equivocal lesion on CT without FDG uptake will usually not be metastatic. PET can also be of help in differentiating hepatic lesions that remain indeterminate by conventional studies. PET may also reveal metastases in sites that escape attention in conventional staging ( Fig. 21-4 ), including soft tissue lesions, retroperitoneal lymph nodes, barely palpable supraclavicular nodes, and painless bone lesions. Exclusion of malignancy requires caution when smaller lesions (<1 cm) are present (see Table 21-1 ). A particular example is a small contralateral lung nodule)—a common finding in the era of chest multislice CT imaging—where negative PET/CT results often do not guarantee certainty, so that invasive sampling (e.g., thoracoscopy) is still needed to exclude malignancy.
PET/CT has long been used to assess pleural involvement, initially with promising results, but recently with more variable results. Small pleural deposits can be missed on PET/CT, because of their low tumor load and/or partial volume effects, whereas false-positive findings may be caused by inflammatory pleural lesions. If the diagnosis of pleural abnormalities will determine the chance for treatment with curative intent, verification of a pathologic process with cytologic analysis or thoracoscopy is often needed.
Influence on Treatment Choices and Planning with Curative Intent
PET has a significant complementary role to CT for two reasons. First, PET can detect unexpected lymph node involvement or distant organ metastatic spread (see Fig. 21-4 ). After a negative conventional staging, previously unknown metastases are found on PET/CT in 5% to 20% of patients, in increasing numbers from clinical stages I–III tumors. Second, PET is able to determine the nature of some lesions that are equivocal on conventional imaging. There is no problem of interpretation when whole-body PET shows metastases in many sites, but an isolated suspect lesion that determines radical treatment intent should always be verified by other tests or tissue sampling, because of the risk for a false-positive finding (see Table 21-1 ) or a second primary tumor. In one large retrospective series, solitary extrathoracic lesions were documented in about 20% of the patients; about half of these were metastatic, whereas the other half were either unrelated to lung cancer (inflammatory or other benign lesions) or second primary tumors.
The effect of adding PET or PET/CT to a standard staging algorithm has been investigated in several randomized controlled trials. Two earlier trials looked at the advantages of stand-alone PET and reported seemingly contradictory results, perhaps because of differences in trial design. In the Dutch trial, which found value in adding PET to the standard workup, the end point—“futile thoracotomy”—was clearly defined as indicating benign disease, explorative thoracotomy, pathologic stage IIIA-N2/IIIB, or postoperative relapse or death within 12 months. In contrast, in the Australian study, which found less value in adding PET, there were no benign lesions, surgery was considered to be of use in some stage IIIA-N2 patients, and no strict follow-up terms were predefined. The Australian trial also focused on clinical stage I and II patients only, from which less additional benefit of PET was expected based on previous nonrandomized accuracy studies.
Three later trials used PET/CT imaging in addition to standard workup, two of which took place in the surgical setting ( Table 21-2 ). The study of Fischer et al largely reproduced the Dutch experience, in which addition of PET led to a significant reduction of futile thoracotomies. The study of Maziak et al mainly looked at improved correct upstaging in resectable stages I–III non–small cell lung cancer (NSCLC) and met this primary end point. Overall, in the two studies there was a 4% to 11% increased detection rate of stage IV disease, and the use of PET/CT led to a change in patient management, both in intent (curative versus palliative) and modality (chemotherapy versus other modalities). In a study of unresectable stage III NSCLC, 21/140 (15%) patients were correctly upstaged with PET/CT versus 4/149 (2.7%) with CT alone. Thus the overall evidence points to significantly more accurate TNM staging with PET/CT than with conventional imaging alone. This leads to true benefits, such as stage migration, better treatment choices, and perhaps better outcome, although the latter still needs to be proven, because randomized controlled trials have been underpowered to assess this end point.
| Study and year | N | Population | Proportion Stages I–II | Primary Outcome |
|---|---|---|---|---|
| Fischer et al 2009 | 189 | Stages I–III | 33% | Futile thoracotomy 52% vs. 35% ( P = 0.05) |
| Maziak et al 2009 | 337 | Stages I–IIIA | 90% | Correct upstaging 6.8% vs. 13.8% ( P = 0.046) |
| Ung et al 2009 | 304 | Stage III | 0% | Correct upstaging 2.7% vs. 15% ( P = 0.002) |
It has been shown in many radiotherapy planning studies that PET/CT influences the accurate delineation of target volumes for radiotherapy. In general the PET-based volume delineations are smaller than those with CT alone, mainly due to more accurate nodal staging ; the lower tumor volume permitted radiation dose escalation to the tumor in a substantial number of patients. Prospective clinical trials using selective nodal irradiation based on PET/CT scanning reported isolated nodal failures in fewer than 5% of patients treated with chemoradiotherapy, which is lower than the 13% rate of false-negative PET results reported in CT-positive lymph nodes. The lower recurrence rate than expected might be explained by the incidental irradiation of lymph nodes adjacent to the planning target volume. In addition, PET/CT-based delineation might be crucial to avoid geographic misses leading to treatment failures. Because of the possibility of false-positive lymph nodes on PET, invasive nodal staging using endosonography (endobronchial ultrasonography or endoscopic ultrasonography) or mediastinoscopy may be warranted if the nodes concerned would have a major impact on defining the radiation treatment field (see Chapter 22 ).
PET-based delineation of the primary tumor usually does not add significantly to that of CT-based delineation, except in situations with postobstructive atelectasis. The optimal method of delineation still remains to be defined. An automated PET/CT delineation may reduce the interobserver variability in treatment planning compared to CT alone. PET may also identify high FDG-uptake regions within the primary tumor as being more radioresistant. Work is in progress to plan higher radiation doses to these potentially radioresistant areas. Radiation dose escalation using an integrated boost to high FDG-uptake regions within the primary tumor proved to be safe and feasible in a small randomized phase II study.
Prognosis
Several PET staging studies have clearly demonstrated TNM stage migration. The possible effect of stage migration may in part account for an apparent improvement in survival of treated patients in both early and advanced disease stage cohorts. The artifactual improvement seen with stage migration is widely referred to as the “Will Rogers phenomenon,” in which patients that move from one stage to another can improve the apparent survival in both stages. (Will Rogers, the American comedian, observed: “When the Okies left Oklahoma and moved to California, they raised the average intelligence level in both states.”) As mentioned earlier, the randomized controlled trials to date have been underpowered to evaluate the potential individual patient survival benefit due to TNM stage migration brought about by PET.
PET has also been shown to predict the prognosis of patients with NSCLC. In a recent systematic review and meta-analysis, the SUV of the primary tumor at diagnosis was found to predict outcome in NSCLC, especially in the earlier stages. These studies almost uniformly found a better overall survival among patients with a metabolic activity lower than the threshold SUV value, calculated from either the most discriminative log-rank SUV value or the median SUV. However, although SUV may be a way to assess prognosis, there is no true cutoff point suitable for broad clinical use. Instead of a true cutoff point, there may rather be a continuous SUV spectrum of a gradually worsening prognosis. When baseline SUV was incorporated as a continuous variable in a Cox proportional hazards model, a one-unit increase in SUV was associated with a 7% increase in hazard of death in resected stages I–III NSCLC and a 6% increase in hazard of death in inoperable NSCLC patients treated with radiotherapy.
Health Economics
As respiratory oncologists, we aim for the best-quality health care for our patients but acknowledge the need for financial prudence. The major cost of modern oncology practice, however, does not lie in the baseline diagnostic process, but in the delivery of expensive treatments and the morbidity related to possible side effects. Therefore application of economic modeling to the use of PET has to be based on both diagnostic and therapeutic aspects of health care expenditure within a daily clinical setting.
In a recent overview of all economic evaluations of PET in oncology performed between 2005 and 2010, it was concluded that the strongest evidence for cost-effective use of PET was for the staging of NSCLC, where there may be benefits both for patients in terms of a possible increase in life expectancy and for the health care system in terms of cost savings resulting from the number of invasive procedures avoided. Taking into account the superior accuracy of PET/CT compared to PET alone in lung cancer staging, the health economic impact in terms of cost-effectiveness can most probably be extended to PET/CT. Since the introduction of PET/CT technology into clinical medicine in 2001, additional studies in respiratory oncology have confirmed the cost-effectiveness of this integrated scanning method. Furthermore, PET has been shown to be cost-effective for characterizing SPNs and to be the most cost-effective diagnostic strategy for nodules of low to moderate pretest probability of malignancy on CT.
Less Classic Indications
Small Cell Lung Cancer
Small cell lung cancer (SCLC) typically shows very high FDG accumulation; however, the value of PET for SCLC is not as well established as for NSCLC. For management of SCLC, the data on the use of PET are less robust than for NSCLC, because the emphasis on systemic and radiation therapy as opposed to surgical resection provides less histologic data to serve as the gold standard. Furthermore, most studies were rather small (mean n = 40) and retrospective in nature. A review of 14 studies comparing FDG-PET with conventional staging procedures found an overall cumulative staging concordance between PET and conventional imaging in 84%. Based on PET, limited stage SCLC was upstaged to extensive stage SCLC in 18% of patients, and extensive stage SCLC was downstaged to limited stage in 11%. The information on PET might result in considerable changes in patient management, ranging from 27% to 47% across studies. The use of PET/CT resulted in changes to the three-dimensional conformal radiation therapy plan in 58% of patients, mainly by decreasing the target volume (in the setting of atelectasis) or detecting unsuspected nodal or pulmonary foci. For prognostic predictions, pretreatment PET values had no value for stages I–III SCLC, but a complete metabolic response on posttreatment PET/CT was associated with better outcome in retrospective analyses.
Mesothelioma
Integrated PET/CT imaging is playing an increasing role in the assessment of suspected or known malignant pleural mesothelioma (MPM, Fig. 21-5 ). PET/CT could be an effective tool in the correct differentiation of malignant (mainly MPM) and benign pleural diseases in asbestos-related CT findings, with an overall accuracy of higher than 90% and a negative predictive value of above 90%. Compared to CT alone, PET/CT is significantly more accurate in baseline TNM staging of patients who are considered appropriate candidates for multimodal therapy. Although PET/CT does not provide additional information about the primary tumor compared to CT alone, it identifies a higher number of metastatic mediastinal lymph nodes and/or unknown distant metastatic disease in up to two thirds of patients, with a significant clinical impact on treatment planning. Early evidence also suggests that PET/CT may have a role in evaluating response to therapy in MPM, which is interesting because the assessment of response in patients with MPM according to standard response criteria on CT is far from simple. More work to define response criteria for MPM on PET is needed. Furthermore, a prospective study in patients with nonsarcomatoid MPM observed that baseline total glycolytic volume on PET was more predictive of survival than CT-assessed TNM stage in a multivariate analysis. These observations of prognostic capability still require prospective validation.
