| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 1995;60:1573-1581
© 1995 The Society of Thoracic Surgeons
Northern California PET Imaging Center and Radiological Associates of Sacramento, Sacramento, California
 |
Abstract |
|---|
 TopFootnotes Abstract IntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Methods. A prospective evaluation of regional and whole-body positron emission tomography (PET) imaging for staging lung cancer was carried out in 99 patients. Mediastinal PET and computed tomography findings were compared with results of surgical staging in 76 patients. Those PET and computed tomography findings that indicated possible distant metastasis were compared with biopsy results and the results of clinical and imaging follow-up.
Results. Sensitivity and specificity for the diagnosis of N2 disease were 83% and 94% for PET and 63% and 73% for computed tomography, respectively. Positron emission tomography showed previously unsuspected distant metastasis in 11 patients (11%), with no demonstrated false-positive results. Normal PET findings were obtained at distant sites of computed tomography abnormality in 19 patients (19%). Clinical and imaging follow-up in 14 of these patients showed no evidence of metastasis. In 1 case, the PET result proved to be falsely negative.
Conclusions. Imaging with PET was more accurate than computed tomography for diagnosis of mediastinal and distant metastasis. Detection of unsuspected metastatic disease by PET may permit reduction in the number of thoracotomies performed for nonresectable disease.
|
Introduction |
|---|
|
TopFootnotesAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
A need exists for an accurate, noninvasive means of staging non-small cell lung cancer (NSCLC) that would permit appropriate patient management without the morbidity and cost of surgical staging. Computed tomography (CT) is the imaging modality used most commonly for preoperative mediastinal staging, but this technique is too inaccurate to provide more than an adjunct to surgical assessment [1–5]. Magnetic resonance imaging has also been evaluated for this purpose, with similarly disappointing results [3, 4]. Both CT and magnetic resonance imaging are anatomic imaging techniques that depend on demonstration of lymph node enlargement for diagnosis of metastasis.
Positron emission tomography (PET) with F-18 fluorodeoxyglucose (FDG) provides functional images of glucose uptake and phosphorylation [6]. Use of FDG PET has the potential to image tumor foci in normal-sized lymph nodes by demonstrating metabolic differences between tumor and normal tissue. A preliminary study of PET in mediastinal staging of NSCLC demonstrated higher sensitivity and specificity than CT [7]. This study was performed without whole-body imaging, so that findings were limited to the thorax and the issue of distant metastasis was not addressed. For a more complete evaluation of the role of PET in lung cancer, we conducted a prospective study of distant as well as mediastinal staging using attenuation-corrected regional imaging and noncorrected whole-body imaging in all patients. Mediastinal PET findings were compared with histologic results, and distant abnormalities were correlated with final diagnosis obtained by biopsy or by imaging and clinical follow-up.
|
Material and Methods |
|---|
|
TopFootnotesAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Positron Emission Tomographic Imaging
Imaging was performed with an ECAT EXACT 921 PET system, manufactured by CTI (Knoxville, TN) and distributed by Siemens Medical Systems (Iselin, NJ). This system has an axial field of view of 16.2 cm with intrinsic transverse resolution of 6.0 mm and axial resolution of 5.5 mm, full width at half maximum. Direct and cross-plane data obtained from 24 detector rings are used to reconstruct 47 image sections, separated by 3.4 mm center to center. When images are reconstructed using a Hann filter with 0.4 cutoff, image resolution is approximately 8 to 10 mm full width at half maximum. Data acquisition, image reconstruction, and image display were performed using standard system software. Patients fasted for at least 4 hours before intravenous injection of 0.143 mCi/kg of FDG, for an average dose of 10 mCi. Nonattenuation-corrected whole-body imaging at 4 minutes per bed position was begun 30 minutes after injection. At the conclusion of whole-body imaging, two static 20-minute emission acquisitions were performed with the imaging volume centered first on the mediastinum and then on the liver. Transmission data were acquired for 10 to 20 minutes, depending on the activity of transmission sources. In the first 42 patients, transmission imaging was performed before tracer injection. In subsequent patients, transmission acquisition was performed at the end of the emission studies.
Computed Tomographic Imaging
Computed tomography had been performed before the PET study in all cases, at one of six different imaging centers. The scanners in use at these sites were all GE 8900 or GE HiLight Advantage systems (GE Medical Systems, Milwaukee, WI) during approximately the first 16 months of the study. For the last 12 months, GE HiSpeed Advantage systems were used at all sites. One-centimeter contiguous image sections were obtained in 2 to 4 seconds with the earlier scanners, and 7-mm sections were obtained in 1 second with the current high-speed systems. Sixty-two studies were performed after intravenous injection of 100 to 200 mL of contrast material. Thirty-seven studies were performed without contrast injection. Standard scanning indices were used. Fifty-two of the 99 patients had also undergone contrast-enhanced CT imaging of the abdomen and pelvis.
Positron Emission Tomography Image Interpretation
All PET images were interpreted at the time of the study by one or two nuclear medicine physicians (P.E.V., T.R.P., M.K.H., or R.W.M.). Images were viewed in axial, coronal, and sagittal planes using an interactive video display system. When CT images of the chest were available, these were used for localization and measurement of the primary tumor and other pulmonary opacities. Mediastinal CT findings were disregarded. No other images were available at the time of interpretation. Overlays of emission and transmission PET images were used to localize metabolically active peripheral lung lesions in some cases. Positron emission tomography images of the chest and abdomen and whole-body images were read as positive or negative for metastatic tumor. In mediastinal staging, we made no attempt to localize visualized activity to a particular nodal station, except for differentiating right from left and hilar from mediastinal activity. For distant metastases, the anatomic site of disease was noted. An involved organ system was treated as a single site, regardless of the number of individual lesions. Thus, skeletal metastasis was counted as a single metastatic site, regardless of the number of bones involved.
Standardized uptake values (SUVs) were determined for primary lesions that were 2 cm or more in diameter. The SUV is a semiquantitative measure of uptake that is obtained by dividing maximum activity detected in the lesion (µCi/mL) by injected dose corrected for body weight (µCi/g). We did not calculate SUVs for mediastinal lesions because these were usually less than 2 cm in diameter, which is the size limit for obtaining accurate activity measurements with the ECAT EXACT 921 system (unpublished data). This size limit was determined by measuring recovery coefficients for spheres of 4 to 38 mm in diameter, containing a standard concentration of Ga-68 [8].
All mediastinal images were reread at the conclusion of the study by two of us (P.E.V., M.K.H.) to evaluate the effect of increase in experience in PET interpretation during the study. Images were reread independently by the two investigators without knowledge of patient identity, clinical data, CT findings, histologic data, or the results of the initial readings. Readings were subsequently compared, and consensus was reached by discussion. At rereading, mediastinal images were graded on a three-point visual scale: 0 = activity less than or equal to mediastinal blood pool activity; 1+ = activity slightly but definitely above mediastinal activity; and 2+ = activity markedly above mediastinal activity. A score of 1+ or 2+ was considered positive for metastasis.
Computed Tomographic Image Interpretation
Computed tomography had been performed before PET referral at one of six different imaging facilities. To ensure uniformity of interpretation, all chest CTs were reinterpreted for mediastinal staging by two experienced readers (G.A.H., H.B.G.). Lymph nodes that measured more than 1 cm in the short axis were considered positive. Each study was read as positive or negative for mediastinal metastasis, without specifying nodal station. Readers were blinded to the clinical data, PET findings, and histologic diagnosis.
Histologic Diagnosis of Mediastinal Metastasis
Tissue for histologic diagnosis was obtained at thoracotomy or mediastinoscopy. The side of the mediastinum that was explored was assessed as involved or uninvolved, according to the histologic diagnosis obtained from examination of permanent sections of lymph node specimens. Midline nodes were considered ipsilateral to the primary tumor. In 4 patients who did not undergo surgical evaluation, diagnosis of metastasis was made on the basis of progression of abnormality on repeat CT imaging.
Diagnosis of Distant Metastasis
Sites of suspected distant metastasis were evaluated by biopsy in 6 patients. In nine cases in which biopsy was not feasible, sites that were positive by PET or CT were evaluated by follow-up CT imaging to demonstrate growth or stability of the imaging abnormalities. In 10 patients with CT abnormalities of the liver, adrenal glands, or contralateral lung, absence of clinical disease 6 to 28 months after imaging was accepted as evidence against metastasis.
Statistical Analysis
Sensitivity, specificity, accuracy, and positive and negative predictive values were calculated for PET and CT diagnosis of mediastinal metastasis. Sensitivity and specificity of PET and CT were compared by the McNemar test [9]. Mean value and standard deviation of the SUVs of the primary tumors were calculated.
|
Results |
|---|
|
TopFootnotesAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Comparison of PET and CT images of the primary tumors and other lung nodules demonstrated a relative lack of anatomic information in the metabolic images. There was little visible variation in FDG uptake between different normal tissues, so that muscle and bone could not be differentiated readily in FDG images. Also, the lung-chest wall and lung-liver interfaces were less well defined in PET than in CT images, and it was difficult to determine from the emission images alone the precise location of small lung lesions that were situated very peripherally. More exact localization was obtained by viewing overlays of emission and transmission PET images.
Mediastinal Staging
In all patients, PET imaging was performed after CT imaging, with a mean time interval of 23 days and a range of 1 to 51 days. Validation of imaging findings was obtained for 76 mediastinal evaluations in 74 patients. Nodal (N) stage was determined histologically in 72 evaluations. Thoracotomy with ipsilateral node sampling was performed in 62 patients. Operative evaluation of both sides of the mediastinum was performed in 2 other patients with bilateral primary tumors, and these were both considered as two mediastinal evaluations. Six patients with positive lymph node findings were evaluated by mediastinoscopy alone.
Forty-nine of the 64 thoracotomies were performed by one of us (D.M.H.). In these cases, the mediastinal and pulmonary nodal stations that were resected were recorded according to the American Thoracic Society classification of nodal locations [10]. Only ipsilateral nodal sampling was performed in patients with unilateral tumors. On the right, this always included stations 4, 7, 8, and 10, as well as the intrapulmonary nodes. On the left, this routinely included stations 5, 8, and 10, and commonly extended to 6 and 7, as well as the intrapulmonary nodes. In the remaining 15 thoracotomies, the extent of lymph node evaluation was more variable.
In 3 patients who had mediastinal metastasis by PET and CT, a reference diagnosis of mediastinal involvement was made without histologic proof, on the basis of subsequent progression of mediastinal lesions on repeat CT imaging. In another patient, mediastinal metastasis was diagnosed on the basis of PET and CT abnormalities, with subsequent shrinkage of the mediastinal abnormalities and an enlarged adrenal gland after chemotherapy.
No validation of mediastinal findings was obtained in 25 patients (25%), and their studies were not considered in the statistical analysis. Fourteen of 18 patients who had PET evidence of distant metastasis were treated without histologic determination of nodal status. Eight other patients with extensive mediastinal disease on imaging were treated on the basis of imaging findings, without independent diagnosis. Three patients with no imaging evidence of nodal disease underwent operations without nodal resection.
Mediastinal metastasis (N2 nodal stage) was demonstrated in 24 of 76 evaluations (32%), histologically in 20 and by progression of imaging abnormalities in four cases. In a further seven cases (9%), only hilar or peribronchial disease (N1 nodal stage) was found. The PET study was positive in 20 of 24 cases (83%) of mediastinal metastasis and in four of seven cases (57%) of metastasis restricted to the hilum or peribronchial nodes. The CT findings were positive in 15 of 24 cases (63%) of mediastinal disease and in one of seven cases (14%) of limited peribronchial or hilar disease. Absence of mediastinal metastasis was demonstrated histologically in 52 of 76 cases (68%). The PET study was normal in 49 (94%) of these cases, and CT was normal in 38 (73%). The positive and negative predictive values for mediastinal disease were 88% and 92%, respectively, for PET and 54% and 79% for CT. Evaluation by the McNemar test showed PET to be more sensitive and specific than CT, with combined p < 0.01. Accuracy for PET and CT was 91% and 70%, respectively.
The results of PET correctly changed the N stage as determined by CT in 18 staging evaluations (24%). The stage was changed from N0 to N2 in 5 patients and from N0 to N1 in 3 patients. In 11 patients, PET results changed the stage from N2 to N0 but in one of these cases, the CT result proved to be correct. Detailed correlation of pathologic and CT findings in this patient indicated that the enlarged nodes were distant from the normal-size node that was histologically involved. In all other cases in which the PET and CT findings differed, the PET finding proved to be correct.
In 2 patients, PET showed evidence of contralateral mediastinal metastasis, whereas CT showed only N2 disease. The ipsilateral mediastinum was histologically positive in these patients, but the contralateral mediastinum was not evaluated histologically, so that the N3 disease was not confirmed. In 1 patient, PET demonstrated supraclavicular nodal metastasis, but this also was not evaluated pathologically. These unconfirmed nodal findings were not included in the statistical analysis.
Discordant PET and CT images of the mediastinum are shown in Figure 1
. In Figure 1A, CT shows normal-size lymph nodes and PET shows intense FDG uptake in one node. Histologic examination demonstrated tumor. In Figure 1B, CT shows a 15-mm preaortic lymph node, with normal FDG uptake on PET. All nodes were found to be histologically negative in this patient.
|
There were three false-positive mediastinal PET studies. In all three cases, CT also was positive. In one case, pathologic examination showed enlarged anthracotic lymph nodes at the site of false-positive uptake, but such nodes were also seen in patients with no abnormal uptake. In the other two patients, only lymph node hyperplasia was seen. This was also a common finding in negative cases.
When PET images were reinterpreted by two readers at the conclusion of the study, without clinical, CT, or histologic information, one study that had been read as 0 was reread as 1+. This proved to be a false-positive reading. Otherwise, the blinded retrospective readings agreed with the blinded prospective readings.
In two of the three false-positive cases, the degree of abnormal uptake at the falsely positive site was mild (1+) rather than intense (2+). In the third, a 2+ abnormality was seen. All other mediastinal sites of 2+ activity proved to be true positive. Mild 1+ abnormality was seen at only two true-positive metastatic sites, so that the predictive value of this finding was 50%. Interpreting mildly elevated uptake as negative for metastasis reduced the sensitivity to 76%, while increasing specificity to 98% and positive predictive value to 94%.
Distant Metastasis
Eleven patients (11%) had evidence of previously unsuspected distant metastatic disease on whole-body PET imaging. Three patients had PET evidence of CT-negative liver metastases, which were confirmed by biopsy in 1 patient (Fig 2) and by later development of a corresponding CT abnormality in another patient. The third patient had PET evidence of skeletal metastasis as well and died of disseminated metastatic disease 26 days after the PET study. Pre-PET CT scans in this patient showed only the primary tumor.
|
Biopsy-confirmed PET lesions in muscle were found in 2 patients, one in the right flank and the other in the anterior abdominal wall. These patients had not undergone previous CT imaging of the abdomen. Another patient had an upper abdominal focus on PET, which was initially CT negative (Fig 3). A second CT study 1 month after the PET study did demonstrate the lesion, which was situated in the periportal region. One patient had PET evidence of adrenal disease with normal CT findings. No follow-up was obtained in this case. Three of the 11 patients in whom PET demonstrated unexpected distant metastasis had no evidence of mediastinal spread.
|
), and in 1, a follow-up CT study of the liver was positive. One patient had a solitary lesion in the contralateral lung on PET and CT that subsequently showed growth on repeat CT imaging. In 1 patient with biopsy-positive liver metastasis, CT showed evidence of one liver lesion, whereas PET showed four lesions. In 4 of 7 patients who had both PET and CT abnormalities at distant sites, PET showed more widespread disease than CT.
|
False-positive distant CT findings were seen in 19 patients (19%). There were more false-positive than true-positive CT abnormalities at distant sites. Eight patients showed adrenal gland enlargement or adrenal masses measuring 2 to 4 cm in diameter, with no abnormality of FDG uptake. In 2 of these patients, repeat CT imaging 10 and 11 months later showed no size increase. Four patients were clinically free of disease 7, 10, 16, and 26 months, respectively, after imaging. No further information was available in 2 patients. Five patients had PET-negative CT abnormalities of the liver, and 4 were clinically free of disease 8, 10, 12, and 22 months later, respectively. No follow-up information was available in the fifth patient.
Six patients had nodules in the contralateral lung on CT imaging that were negative by PET. In one case, increase in lesion size on repeat CT imaging 5 months later indicated that the PET result represented a false-negative finding. This lesion was 5 mm in diameter at the initial PET examination. In 1 patient, the nodule was resected and was found to be benign at histologic examination, and in another patient, the nodule had resolved on follow-up chest radiography 7 months later. Two other patients had no clinical evidence of tumor at 6 and 8 months. No follow-up information was available in 1 patient. Overall, there was direct or indirect evidence supporting a benign diagnosis of the PET-negative CT lesions in 14 of the 19 patients. In one lung lesion, the PET result proved to be falsely negative.
|
Comment |
|---|
|
TopFootnotesAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Evidence of unsuspected distant metastasis on PET was found in 11 of 99 patients (11%), with pathologic or follow-up confirmation of the diagnosis in all eight cases in which confirmatory data were available. Eight of these patients had undergone CT imaging that included one or more sites of metastasis that were positive by PET. Seven other patients (7%) had adrenal, liver, or lung lesions by CT as well as PET imaging, but the CT findings alone were of uncertain significance because false-positive findings at these sites were encountered in 19 patients (19%). The PET finding of hypermetabolism at the sites of CT abnormality had a major impact on the perceived significance of the CT results.
In mediastinal staging, PET commonly showed evidence of tumor in nodes that were not enlarged by CT and the absence of tumor in nodes that were enlarged by CT. In 16 of 17 discordant cases, the PET result proved to be correct. In 1 patient with histologic evidence of mediastinal involvement, CT showed lymph node enlargement and PET was negative. However, detailed correlation of CT and histologic findings showed that the enlarged nodes were not histologically positive and the histologically positive node was not enlarged, indicating a coincidence of false-negative and false-positive results. Fourteen of 29 positive CT results proved to be falsely positive.
Positron emission tomography imaging was more sensitive than CT for mediastinal staging because of its ability to demonstrate tumor metabolism in normal-size lymph nodes. However, PET was also limited by minimum detectable tumor mass. Thus, PET detected tumor deposits in nodes that were not enlarged in 3 patients, but failed to detect microscopic tumor in 4 other patients. Also, a 5-mm lung lesion that was negative by PET proved to be metastatic on the basis of lesion growth demonstrated by CT imaging.
Detectability of tumor by PET depends primarily on two factors: tumor uptake of FDG and tumor size. Variation in detectability on the basis of uptake was to be expected, as the SUVs of primary lesions that were more than 2 cm in diameter varied from 3.2 to 22.3. Variation according to size was also to be expected because the recovery coefficient of the tomograph, using a Hann filter with 0.4 cutoff for image reconstruction, decreased rapidly as object diameter decreased below 2 cm (unpublished data). This finding is consistent with the observations of Hoffman and associates [8] and Kessler and co-workers [13] in relating recovery coefficient to object size and instrument resolution. Detection by PET of tumor foci in normal-size lymph nodes implied very high tumor uptake, with tumor to normal tissue ratios of 10:1 or greater. These findings are consistent with early in vitro evaluations in animal tumor models reported by Wahl and colleagues [14].
The high tumor to normal tissue uptake ratios that were achieved with FDG were also reflected in highly reproducible interpretation of mediastinal PET images. Blinded rereading of mediastinal images at the conclusion of the study showed 99% concordance between the prospective and retrospective readings. The high tumor uptake also permitted interpretation of image results on a three-point, 0 to 2+ scale, with no provision for ``possibly abnormal'' readings.
The range of PET image abnormalities that may be encountered in tumor-free lymph nodes, resulting from benign causes, remains to be determined. Because of the high specificity of FDG PET in NSCLC, more pathologically correlated studies are needed before a meaningful experience of false-positive results can be accumulated.
In our three cases of false-positive mediastinal PET findings, no specific pathologic lymph node changes were found to explain the metabolic hyperactivity. In the evaluation of undiagnosed solitary pulmonary nodules, hypermetabolism has been described in inflammatory processes and in granulomatous lesions, including histoplasmosis, aspergillosis, and active tuberculosis, as well as in carcinoma [15–17]. These pathologic conditions were not encountered in the present study, probably reflecting the higher prevalence of these diseases in patients with undiagnosed lung nodules compared with patients with diagnosed lung cancer. Because the prevalence of nonpyogenic pulmonary infections varies among geographic regions, causes of false-positive PET findings that are encountered in staging lung cancer may also vary geographically.
The present findings suggest that it may be appropriate to interpret PET studies of the mediastinum for high specificity or high sensitivity, depending on the intended clinical use of the result. Interpreting 1+ studies as negative for tumor increased the specificity to 98% at a small cost in sensitivity. Such high-specificity reading would be appropriate if a positive PET result were to be used to exclude attempted curative surgery, without histologic confirmation. On the other hand, if a positive PET result were to be used as an indication for prethoracotomy mediastinoscopy, high-sensitivity reading would be appropriate. An analysis by Malenka and associates [18] indicated that a noninvasive technique for mediastinal staging would have the same life-expectancy outcome as strategies involving surgical staging if specificity was 90% or higher; FDG-PET may reach this goal.
The greater sensitivity of PET in detecting distant soft-tissue lesions such as liver metastases also resulted from the functional nature of the PET technique. Computed tomography depends upon differences in the structural characteristics of tissue, manifested as differential absorption of photons. If there is no difference in photon absorption between tumor and normal tissue, the lesion remains undetected. The present results indicate that the difference in glucose uptake between lung cancer and normal tissue in PET images is more visible than the difference in photon attenuation in CT images. The exception appears to be the lung, where the CT contrast between tumor and air-containing lung is high. In our patients, CT detected smaller lung lesions than PET.
Calculation of SUV permits more objective classification of lesion activity than visual assessment alone and permits semiquantitative comparison of FDG uptake between patients and between different tumor types. However, Lowe and colleagues [19] found that SUV determination and visual assessment were equally accurate for differentiating benign from malignant pulmonary nodules. Our findings support their conclusions. Determination of SUV may increase the sense of diagnostic objectivity, but seemed to add little to diagnostic accuracy.
Precise anatomic localization of some PET abnormalities was difficult because of the paucity of anatomic information in metabolic images. The difference in tracer uptake between the liver and adjacent lung was clearly apparent, but the interface between the two organs was difficult to localize precisely, so that it was not possible to differentiate a dome-of-liver lesion from an inferior lung lesion in one case. Similarly, a peripheral lung lesion was hard to differentiate from a pleural or chest wall lesion, and individual bone abnormalities could not be distinguished reliably from adjacent soft-tissue abnormalities. Images from CT provided more anatomic detail than PET images and permitted more accurate localization of lesions. Computed tomography also permitted accurate measurements of abnormalities, which could not be obtained from the PET images.
Coregistration of PET and CT images has been used by Wahl and co-workers [7] in staging NSCLC to combine the anatomic information of CT with the functional information of PET, but this did not add to the diagnostic information obtained by visual image correlation. We also found that visual correlation was sufficient for metabolic diagnosis of pulmonary and distant CT abnormalities. Addition of PET findings to CT findings allowed determination of the metabolic and diagnostic significance of anatomic CT abnormalities. Addition of CT findings to PET findings was useful for anatomic localization of metabolic activity in some cases, but did not aid in determining diagnostic significance. Correlation of PET with CT was not needed to determine the significance of PET results. Such correlation was reserved for a small number of studies in which precise anatomic localization was uncertain and clinically important, and the lesion was visible on CT as well as PET images.
In evaluating new diagnostic modalities in oncology, it is desirable to use definitive histologic diagnosis as the only reference test for measuring performance. However, this is not always possible. In the present study, this would have resulted in substantial verification bias [20] because some patients with mediastinal spread and most patients with distant metastasis did not undergo surgery and pathologic verification of mediastinal disease. As a result, patients with negative PET and CT findings would have been overrepresented in the test group. Such selection bias increases the proportion of negative results, both true and false, thereby reducing the measured sensitivity and increasing the measured specificity of the imaging techniques. In addition, in the case of distant metastasis, pathologic verification was obtained only when imaging results were positive. Begg and McNeil [20] suggested that careful follow-up of all patients with unverified disease is needed to reduce verification bias and that analysis of results should include all patients, not only those with pathologic verification. For these reasons, we included the mediastinal results of 4 patients with nonresectable disease, in whom a reference diagnosis was determined by follow-up imaging.
The problem of verification was greatest in the diagnosis of distant metastasis, for which it was not practical to obtain pathologic verification of most lesions and verification was rarely obtained in the absence of imaging abnormality. This problem is inherent to any evaluation of disseminated metastatic disease. Statistical analysis of distant findings in terms of sensitivity and specificity was not possible because metastatic lesions that were missed by PET and CT generally remained undetected, and verification of negative results was obtained only when PET and CT results were discordant. Hence, most distant imaging abnormalities were validated on the basis of disease progression or absence of progression, rather than pathology, and PET and CT results were reported only in comparative terms. Positron emission tomography detected distant metastasis in more patients than did CT, and no distant PET abnormalities proved to be falsely positive.
Case selection bias of a more subtle form may also have an impact on evaluation of PET in lung cancer. When PET is used primarily for the diagnosis of indeterminate pulmonary nodules, with staging as a secondary consideration [21], tumor is likely to be less advanced than in patients who have an established diagnosis of lung cancer at the time of imaging. Patients with small, indeterminate nodules on CT are likely to have fewer and smaller metastases than patients with larger tumors, resulting in a higher false-negative rate and lower measured sensitivity. The majority of patients in the present study had a histologic diagnosis at the time of the PET study.
Verification and case-selection biases would have affected both imaging modalities equally, without biasing the comparison of PET and CT results. Other possible sources of bias that would have favored PET over CT include the time interval between CT and PET imaging, unequal skills in test performance, and bias in test interpretation. As a result of the interval between the CT and PET examinations, the tumors were an average of 23 days more advanced by the time of PET study. This interval between studies reflected community practice, whereby the investigation for lung cancer was usually initiated by the discovery of a nodule on chest radiography, followed by chest CT examination. Imaging by PET was usually not requested until the patient was seen by a thoracic surgeon. Simultaneous performance of PET and CT examinations was not a practical option.
This study was conducted in a PET imaging center. The investigators had no control over the performance of the CT examinations, which were done at one of three outpatient imaging centers or one of three hospitals. State-of-the-art CT equipment and standard protocols were in use at all sites, but injection of contrast was not employed in all cases. Rereading of chest CT results by two of the investigators was used to confirm study quality and to minimize variability in interpretation. Some degree of inequality in performance was likely because the PET operators were intensely involved in the evaluation of a new imaging technology, whereas the CT examinations were performed at multiple imaging sites as routine procedures [20, 22].
A possible source of interpretation bias was the availability of chest CT images at the time of initial PET interpretation in some patients. Chest CT images were used only for localization and measurement of intrapulmonary lesions, and mediastinal findings were disregarded. Extrathoracic imaging studies were not available at PET interpretation. When PET and CT assessments of the mediastinum differed, the PET findings proved to be correct in 18 of 19 cases, so that consideration of CT findings would probably not have provided an advantage. The prospective evaluation was initially undertaken as a field study of PET imaging rather than a fully controlled comparison of the two modalities. Field studies that observe conventional clinical practices are more likely to be directly generalizable than studies in a controlled environment [20]. Control was introduced by the retrospective, fully blinded rereading of PET and CT chest images.
Our findings of 63% sensitivity and 73% specificity for CT in the diagnosis of N2 disease are comparable to those obtained in other recent studies that involved extensive node sampling and large study populations [1–5]. This similarity of findings suggests that the possible sources of bias did not greatly affect our evaluation of mediastinal CT imaging.
On the basis of the present findings, it may be possible to improve significantly the accuracy of nonsurgical staging of lung cancer by replacing abdominal and pelvic CT with whole-body PET imaging. Detection of unsuspected metastasis by PET would reduce the number of thoracotomies performed for nonresectable disease. Chest CT would continue to be needed for accurate determination of tumor T stage and evaluation of thoracic anatomy.
Multiinstitutional trials are needed to establish the reproducibility of these results. After the conclusion of this initial evaluation of accuracy, we are continuing to evaluate the impact of PET findings on management and on the cost of patient care. With the changes that are occurring in the funding of health care, it has become obligatory to seek cost-effective algorithms for the management of malignant disease. Unfortunately, these changes may also make it more difficult to do so, if development of the algorithms involves the evaluation of a new imaging technology. Results obtained from preliminary assessments of PET in oncology justify more extensive prospective trials, as the potential impact of this technology on the clinical and cost-effectiveness of patient management is significant. However, the resources needed for such trials are difficult to procure.
|
Acknowledgments |
|---|
|
TopFootnotesAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
|
Footnotes |
|---|
|
TopFootnotesAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Address reprint requests to Dr Valk, Northern California PET Imaging Center, 3195 Folsom Blvd, Sacramento, CA 95816.
|
References |
|---|
|
TopFootnotesAbstractIntroductionMaterial and MethodsResultsCommentAcknowledgmentsReferences |
|---|
Related Articles
This article has been cited by other articles:
|
|
 |
|
  Y. C. Ung, D. E. Maziak, J. A. Vanderveen, C. A. Smith, K. Gulenchyn, C. Lacchetti, W. K. Evans, and Lung Cancer Disease Site Group of Cancer Care Onta 18Fluorodeoxyglucose Positron Emission Tomography in the Diagnosis and Staging of Lung Cancer: A Systematic Review J Natl Cancer Inst, December 5, 2007; 99(23): 1753 - 1767. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Ebihara, H. Nomori, K. Watanabe, T. Ohtsuka, T. Naruke, K. Uno, I. Kuwahira, and K. Eguchi Characteristics of Advantages of Positron Emission Tomography over Computed Tomography for N-staging in Lung Cancer Patients Jpn. J. Clin. Oncol., November 1, 2006; 36(11): 694 - 698. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. E. Posther, L. M. McCall, D. H. Harpole Jr., C. E. Reed, J. B. Putnam Jr., V. W. Rusch, and B. A. Siegel Yield of Brain 18F-FDG PET in Evaluating Patients with Potentially Operable Non-Small Cell Lung Cancer J. Nucl. Med., October 1, 2006; 47(10): 1607 - 1611. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. F. Munden, S. S. Swisher, C. W. Stevens, and D. J. Stewart Imaging of the Patient with Non-Small Cell Lung Cancer Radiology, December 1, 2005; 237(3): 803 - 818. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. Birim, A. P. Kappetein, and A. J.J.C. Bogers Reply Ann. Thorac. Surg., November 1, 2005; 80(5): 1980 - 1981. [Full Text] [PDF]
|
|
|
|
 |
|
 B. S. Halpern, C. Schiepers, W. A. Weber, T. L. Crawford, B. J. Fueger, M. E. Phelps, and J. Czernin Presurgical Staging of Non-small Cell Lung Cancer: Positron Emission Tomography, Integrated Positron Emission Tomography/CT, and Software Image Fusion Chest, October 1, 2005; 128(4): 2289 - 2297. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. Sachs and T. V. Bilfinger The Impact of Positron Emission Tomography on Clinical Decision Making in a University-Based Multidisciplinary Lung Cancer Practice Chest, August 1, 2005; 128(2): 698 - 703. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
O. Birim, A. P. Kappetein, T. Stijnen, and A. J.J.C. Bogers Meta-Analysis of Positron Emission Tomographic and Computed Tomographic Imaging in Detecting Mediastinal Lymph Node Metastases in Nonsmall Cell Lung Cancer Ann. Thorac. Surg., January 1, 2005; 79(1): 375 - 382. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 L. Schrevens, N. Lorent, C. Dooms, and J. Vansteenkiste The Role of PET Scan in Diagnosis, Staging, and Management of Non-Small Cell Lung Cancer Oncologist, November 1, 2004; 9(6): 633 - 643. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. L. Aquino and A. J. Fischman Does Whole-Body 2-[18F]-Fluoro-2-Deoxy-D-Glucose Positron Emission Tomography Have an Advantage Over Thoracic Positron Emission Tomography for Staging Patients With Lung Cancer? Chest, September 1, 2004; 126(3): 755 - 760. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. J. Lynch, C. D. Wright, N. C. Choi, S. L. Aquino, and E. J. Mark Case 26-2004 - A 56-Year-Old Woman with Cough and a Lung Nodule N. Engl. J. Med., August 19, 2004; 351(8): 809 - 817. [Full Text] [PDF]
|
|
|
|
 |
|
 M. Tomita, Y. Matsuzaki, M. Edagawa, T. Shimizu, M. Hara, and T. Onitsuka Combined Procedures for Mediastinal Staging in Non-small Cell Lung Cancer Asian Cardiovasc Thorac Ann, June 1, 2004; 12(2): 125 - 129. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. C. Detterbeck, S. Falen, M. P. Rivera, J. S. Halle, and M. A. Socinski Seeking a Home for a PET, Part 1: Defining the Appropriate Place for Positron Emission Tomography Imaging in the Diagnosis of Pulmonary Nodules or Masses Chest, June 1, 2004; 125(6): 2294 - 2299. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. J. Schaffler, G. Wolf, H. Schoellnast, R. Groell, A. Maier, F. M. Smolle-Juttner, M. Woltsche, G. Fasching, R. Nicoletti, and R. M. Aigner Non-Small Cell Lung Cancer: Evaluation of Pleural Abnormalities on CT Scans with 18F FDG PET Radiology, June 1, 2004; 231(3): 858 - 865. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Nomori, K. Watanabe, T. Ohtsuka, T. Naruke, K. Suemasu, and K. Uno The size of metastatic foci and lymph nodes yielding false-negative and false-positive lymph node staging with positron emission tomography in patients with lung cancer J. Thorac. Cardiovasc. Surg., April 1, 2004; 127(4): 1087 - 1092. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 A. Sharma, P. Fidias, L. A. Hayman, S. L. Loomis, K. H. Taber, and S. L. Aquino Patterns of Lymphadenopathy in Thoracic Malignancies RadioGraphics, March 1, 2004; 24(2): 419 - 434. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. G. Pfister, D. H. Johnson, C. G. Azzoli, W. Sause, T. J. Smith, S. Baker Jr, J. Olak, D. Stover, J. R. Strawn, A. T. Turrisi, et al. American Society of Clinical Oncology Treatment of Unresectable Non-Small-Cell Lung Cancer Guideline: Update 2003 J. Clin. Oncol., January 15, 2004; 22(2): 330 - 353. [Full Text] [PDF]
|
|
|
|
|
|
J. D. Bradley, C. A. Perez, F. Dehdashti, and B. A. Siegel Implementing Biologic Target Volumes in Radiation Treatment Planning for Non-Small Cell Lung Cancer J. Nucl. Med., January 1, 2004; 45(90010): 96 |
