Abstract
Lung cancer is the leading cause of cancer-related death in industrialized countries. The overall mortality rate for lung cancer is high, and early diagnosis provides the best chance for survival. Diagnostic tests guide lung cancer management decisions, and clinicians increasingly use diagnostic imaging in an effort to improve the management of patients with lung cancer. This systematic review, an expansion of a health technology assessment conducted in 2001 by the Institute for Clinical and Evaluative Sciences, evaluates the accuracy and utility of 18fluorodeoxyglucose positron emission tomography (PET) in the diagnosis and staging of lung cancer. Through a systematic search of the literature, we identified relevant health technology assessments, randomized trials, and meta-analyses published since the earlier review, including 12 evidence summary reports and 15 prospective studies of the diagnostic accuracy of PET. PET appears to have high sensitivity and reasonable specificity for differentiating benign from malignant lesions as small as 1 cm. PET appears superior to computed tomography imaging for mediastinal staging in non-small cell lung cancer (NSCLC). Randomized trials evaluating the utility of PET in potentially resectable NSCLC report conflicting results in terms of the relative reduction in the number of noncurative thoracotomies. PET has not been studied as extensively in patients with small-cell lung cancer, but the available data show that it has good accuracy in staging extensive versus limited-stage disease. Although the current evidence is conflicting, PET may improve results of early-stage lung cancer by identifying patients who have evidence of metastatic disease that is beyond the scope of surgical resection and that is not evident by standard preoperative staging procedures. Further trials are necessary to establish the clinical utility of PET as part of the standard preoperative assessment of early-stage lung cancer.
Introduction
Lung cancer is the leading cause of cancer-related deaths in both men and women in industrialized countries. The overall mortality rate for lung cancer is high, and early diagnosis provides the best chance for survival. Diagnostic tests guide lung cancer management decisions, and diagnostic imaging is increasingly being used in an effort to improve the clinical management of patients with lung cancer.
Whereas traditional radiologic imaging technologies (e.g., computed tomography [CT] scan, magnetic resonance imaging) provide structural information and define disease states on the basis of gross anatomical changes, 18fluorodeoxyglucose positron emission tomography (PET) imaging provides information on the biochemical processes that may precede gross anatomic change. PET imaging is potentially useful in tumor imaging in particular due to the uptake of the radiolabeled glucose analog 18fluorodeoxyglucose (18FDG) by tumor tissue as a result of increased glycolysis in some cancers as compared with most normal tissue.[1,2] This increased glycolysis has been linked to both an increase in the number of glucose membrane transporters and an increase in the activity of the principal enzymes that control the glycolytic pathways.[3]
Two main types of instrumentation have been used for PET imaging: dedicated PET scanners and gamma cameras that have been modified for coincidence imaging.[4-8] Dedicated PET scanners consist of multiple detectors that are arranged in a partial or full ring; the detection sensitivity of a partial ring scanner is less than that of a full ring scanner. Gamma-camera coincidence imaging uses a modified two- or three-headed gamma camera that rotates around the patient. These units are less expensive than dedicated PET, but they are also less sensitive and are no longer being manufactured. They do, however, remain in use in some institutions and are the subject of some of the literature on PET. Recent advances in imaging technology combine PET and CT to provide both functional and anatomical information simultaneously, thus improving localization accuracy.[9,10]
PET data may be analyzed qualitatively, semiquantitatively, or fully quantitatively. Qualitative visual interpretation of PET data involves the subjective assessment of differences in contrast and requires only a static emission scan, with or without a transmission scan and attenuation correction.[4-10] This analytic approach is particularly useful in assessing substantial changes (e.g., decrease in tumor metabolic activity following therapy or the development of new lesions) but is not as valuable as other approaches in assessing more subtle metabolic changes.[11] Semiquantitative approaches include analyses of tumor to normal tissue ratios and standardized uptake values (SUVs). The SUV is the ratio of activity in tissue per unit volume to the activity in the injected dose per patient body weight and is widely used because of its simplicity. It requires only a static scan with accurate instrument calibration, and it is about as discriminating as fully quantitative methods.[12] A number of fully quantitative (or kinetic) methods are used to measure glucose metabolic rate dynamically, thereby providing more detailed information about tumor response.[8] All three types of methods have strengths and weaknesses, and the optimal approach has yet to be established in prospective trials.[11] A full review of PET scan technology was published in 2004.[13]
Staging with PET has the potential to allow clinicians to define more accurately than with conventional staging methods those patients who have metastatic disease and who would, therefore, not benefit from surgery or other localized therapies. Conventional lung cancer staging procedures are imperfect in their ability to identify those asymptomatic patients with occult metastases for whom surgical intervention would be futile, as manifested by the fact that a substantial proportion of such patients go on to develop metastatic disease shortly after thoracotomy. Moreover, if PET scanning could accurately stage non-small cell lung cancer (NSCLC) and also enable concurrent detection of mediastinal and distant metastases, cervical mediastinoscopy or other imaging studies that are presently required to evaluate the mediastinum in patients with NSCLC might be avoided.
PET also has the potential to solve the dilemma of the diagnostic workup of a solitary pulmonary nodule (SPN) when the SPN is not amenable to fine needle aspiration biopsy because of size, location, or associated medical comorbidities. Similarly, open biopsy would not need to be risked if the SPN were known to be benign. Finally, the result of a fine needle aspiration biopsy may not be diagnostic of malignancy because it may be falsely negative if the needle was inaccurately placed or scant cellular material was obtained.
PET has the potential to be useful in the staging of small-cell lung cancer (SCLC).[14-16] SCLC is the most aggressive type of lung cancer; tumors are typically fast growing and 60% to 70% of patients present with extensive-stage disease.[17] However, there is less information on PET in the staging of SCLC, and thus uncertainty remains about its utility for distinguishing patients with extensive disease from those with limited disease.
In an initial search for evidence-based reports on PET in the diagnosis and staging of lung cancer, we identified a 2001 report from the Institute for Clinical and Evaluative Sciences (ICES).[18] This systematic review of the PET scanning literature in general covered the period through December 2000; with subsequent updates,[19] the literature review was current to September 2004. The ICES report reviewed the diagnostic accuracy, effect on patient outcomes, and cost-effectiveness of PET based on a systematic review of the peer-reviewed and online PET scanning literature, focusing on the use of dedicated PET scanners. The ICES report provided evidence for the better efficacy of PET than CT in predicting the histologic status of mediastinal lymph nodes and for detecting pleural involvement and malignant pleural effusion in patients with carcinoma of the lung.
The ICES review was regarded as a high-quality review of the evidence and was expanded in the current systematic review to serve as the basis for clinical practice recommendations. In this review, the Lung Cancer Disease Site Group of Cancer Care Ontario's Program in Evidence-Based Care provides updated evidence on the use of PET in lung cancer. The conclusions from the ICES review and the results of primary studies retrieved as part of this updated search related to the accuracy or utility of PET in diagnosis or staging are organized into three sections: diagnosis of SPNs, staging of NSCLC at initial diagnosis, and staging of SCLC. Within each section, we summarize the findings of the ICES review and the primary studies included in the ICES report, followed by a detailed description of the results of the systematic reviews and primary studies retrieved in the updated search. This systematic review and the resulting guideline recommendations are intended to promote evidence-based practice in Ontario, Canada.
Study Identification
Literature Search Strategy
The literature search was conducted in the Cochrane Database of Systematic Reviews (2006, Issue 1), EMBASE (1996 through 2006, week 19), and MEDLINE (1996 through May 2006). In addition, the conference proceedings of the American Society of Clinical Oncology (2004-2005) and Physician Data Query clinical trials database were searched for primary studies. The following subject-specific terms (i.e., MeSH terms in MEDLINE and EMTREE terms in EMBASE) were searched: "Lung carcinoma," "Lung carcinogenesis," "Lung metastasis," "Carcinoma, non-small-cell lung," "Carcinoma, small cell," "Lung neoplasms," "Lung cancer," and the phrase "Non-small cell lung" used as a text word were combined with "Positron emission tomography," "Tomography, emission computed," "fluorodeoxyglucose F18," and both of the following phrases used as text words: "PET" and "Positron emission tomography". These terms were then combined with the search terms for the following publication types: practice guideline, systematic review, biomedical technology assessment, and metaanalysis. The search was restricted to publications in the English language. In addition, Web sites of practice guideline and health technology assessment organizations were searched in late 2004 and early 2005. The Canadian Medical Association Infobase, the National Guidelines Clearing House, and the National Institute for Clinical Excellence were searched on May 13, 2005; the Web site of the Canadian Coordinating Office for Health Technology Assessment was searched on December 23, 2004; and the Centre for Reviews and Dissemination was searched on February 1, 2005.
Study Inclusion Criteria
Evidence-based reports (i.e., health technology assessments, practice guidelines, systematic reviews, and meta-analyses) that evaluated the use of PET in the staging and diagnosis of lung cancer were reviewed for inclusion in this systematic review if they were published in English after 1999. Primary studies (full reports or abstracts) that were published after the period covered in the ICES review or that examined the use of PET in staging SCLC were selected if they met the following criteria: (a) they were randomized or single-arm prospective studies that focused on PET scanning in the diagnosis and/or staging of lung cancer compared with an appropriate reference standard and (b) they reported at least one of the following measures of effectiveness/benefit: PET specificity and or sensitivity, measures of accuracy of staging, changes in patient management, or improvements in patient outcomes (e.g., survival). Studies involving NSCLC were excluded if they had fewer than 35 subjects; there was no sample-size limitation for studies involving SCLC.
Literature Search Results
In addition to the ICES report, 12 additional evidence-based reports were retrieved. The ICES report was the most comprehensive of the reports; therefore, we include in our systematic review only those reports that were meta-analyses or addressed a question not covered by the ICES report.[21-26] The ICES report included 21 prospective, observational studies and 22 randomized controlled trials (RCTs) on the diagnosis of SPN or the staging of primary NSCLC, all of which were included in other evidence-based reports. We also identified 15 prospective studies (including one RCT) that examined PET in the staging and diagnosis of lung cancer that were published after completion of the ICES report (October 2004 or later). Table 1 lists all prospective studies in ICES and those identified and included in this update. We included multiple publications from the same study if each report provided additional relevant data.
Diagnosis of Solitary Pulmonary Nodules
Findings of Institute for Clinical and Evaluative Sciences
The ICES report[18,19] evaluated four prospective studies[27-30] on the role of PET in the diagnosis of SPN ( Table 2 ). The ICES report concluded that PET has proven efficacious in distinguishing benign from malignant SPN and, by reducing the number of unnecessary thoracotomies performed for SPN, would reduce patient morbidity.
Results of Systematic Reviews
Two systematic reviews[23,24] assessed the accuracy of PET in the diagnosis of SPN. Fischer et al.[23] estimated the mean sensitivity and specificity independently for the identification of malignant pulmonary nodules and masses. The mean sensitivities and specificities were 0.96 (standard error [SE] = 0.01) and 0.78 (SE = 0.03), respectively, for dedicated PET and 0.92 (SE = 0.04) and 0.86 (SE = 0.04), respectively, for gamma-camera PET. Sensitivity was statistically significantly lower with gamma-camera PET than with dedicated PET (P<.005). Relative to the ability of PET to discriminate between malignant and benign pulmonary changes, there was no statistically significant difference between the methods of interpretation (SUV, visual, or both) of PET scans. The authors concluded that PET has value for determining if a pulmonary nodule is malignant or benign but recommended that studies be conducted in populations with low prevalence of NSCLC.
The meta-analysis by Gould et al.[24] included 40 studies of pulmonary lesions and used a meta-analytic method to construct summary receiver operating characteristic (SROC) curves. An SROC curve is used to summarize ROC data from multiple studies, i.e., in the context of a meta-analysis.[31] The maximum joint sensitivity and specificity of PET from the SROC curve was 91.2% (95% confidence interval [CI] = 89.1% to 92.9%). In clinical practice, at a median specificity of 78%, the sensitivity of PET from the SROC curve would correspond to 97% because most studies use thresholds that favor sensitivity over specificity. There was no difference in the diagnostic accuracy of PET imaging for pulmonary nodules based on size (P =.43), for semiquantitative versus qualitative methods of analysis (P =.52), or for studies using dedicated PET versus gamma-camera PET (P =.19). Gould et al.[24] concluded that PET has high sensitivity and intermediate specificity for identifying malignant pulmonary nodules and larger mass lesions but that limited data exist for nodules less than 1 cm in diameter.
Results of Primary Studies
Seven prospective studies[20,27-30,32,33] have examined the use of PET to differentiate benign and malignant SPN ( Table 2 ). Most of the seven studies enrolled patients with indeterminate pulmonary lesions on chest radiography and used histopathology results as the reference standard. Sensitivity in these seven studies ranged from 79% to 100%, except for those that used ground-glass opacity images (see below). Specificity was more variable and ranged from 40% to 90%. Croft et al.[28] reported a specificity of 40%; however, their patient population had a high incidence of granulomas, which increased the number of false positives. Nomori et al.[33] also reported a low sensitivity and specificity; however, this study selected nodules on the basis of ground-glass opacity images on CT, which show low uptake of 18FDG on PET imaging. PET data were evaluated independently of the reference standard in six of the studies.[20,27-30,33] In one trial,[29] only some of the study group received confirmation of the diagnosis by the reference standard, which could result in biased estimates of overall diagnostic accuracy. Six of the studies[20,28-30,32,33] clearly specified explicit criteria for defining a positive PET test result. Two studies[20,32] were conducted by the same research group, and it is not clear if the same patients were included in both studies.
Nomori et al.[32] evaluated 151 noncalcified nodules that were less than 3 cm in diameter in 131 patients. Of these, 15 nodules could not be diagnosed as malignant or benign and were excluded from analyses. Among the remaining 136 nodules, PET could not detect abnormal 18FDG activity in the 20 nodules that were less than 1 cm in diameter, of which eight were malignant. PET correctly detected 57 of 63 malignant nodules that were solid on CT but was positive for only one out of 10 malignant nodules with a faint or ground-glass appearance on CT. All of the malignant nodules with ground-glass images on CT were histologically adenocarcinoma.
Another trial by Nomori et al.[20] compared visual and semiquantitative analyses for nodules of 1-3 cm in diameter. In nodules greater than 1 cm, PET was negative or faintly positive in patients with histologically well- or moderately differentiated adenocarcinoma. The study also found no difference in sensitivity and specificity between visual assessment and semiquantitative methods for nodules graded as definitely positive or negative. However, in nodules that were faintly positive, using the contrast ratio to the contralateral lung and contrast ratio to the cerebellum resulted in statistically significantly higher sensitivity than the SUV.
Primary Non-Small Cell Lung Cancer Staging: Utility and Accuracy of
Findings of Institute for Clinical and Evaluative Sciences
The ICES report included 14 prospective studies[34-46] examining the effectiveness of PET in determining the true extent or stage of primary NSCLC ( Table 3 and Table 4 ). According to the ICES report, the evidence on whether preoperative PET would reduce the number of unnecessary thoracotomies for patients diagnosed with lung cancer was "conflicting."
Results of Primary Studies of
18Fluorodeoxyglucose Positron Emission Tomography Utility in Non-Small Cell Lung Cancer StagingThree randomized clinical trials have evaluated the value of preoperative PET assessment for NSCLC,[47-49] two of which[48,49] were included in the ICES report ( Table 5 ). The PLUS (PET in Lung Cancer Staging) multicenter trial[48] randomly assigned 188 patients with suspected lung cancer to either PET followed by further standard diagnostic and therapeutic procedures or standard diagnostic and therapeutic procedures alone. Half of the patients had a definite diagnosis of NSCLC, and 70% of the patients in each group had clinical stage I-II disease at baseline.[50] The primary outcome was the number of futile thoracotomies. Thoracotomy was regarded as futile if the patient had benign disease, exploratory thoracotomy only, pathological stage IIIA (mediastinal node positive) or IIIB disease, or postoperative relapse or death within 12 months of randomization. The addition of PET to conventional workup led to a 51% (95% CI = 32 to 80, P = .003) relative reduction in futile thoracotomies (from 41% in the conventional workup arm to 21% in the conventional plus PET arm) and prevented unnecessary surgery in 20% of patients with suspected NSCLC. Twenty-seven percent of patients in the combined PET and conventional workup were reclassified as a higher stage as a result of the PET evaluation, compared with 12% of patients in the conventional workup group.
An Australian multicenter trial[49] randomly assigned 183 patients with histologically or cytologically proven stage I-II NSCLC to conventional workup, either with or without PET imaging.[49] The primary endpoint was the proportion of patients undergoing thoracotomy. PET led to reclassification to a higher stage of 24 patients and confirmed staging in 61 patients. There was no statistically significant difference between the trial arms in the number of thoracotomies performed (P = .2), and PET resulted in changes in patient management in only 14% of patients. PET could have altered patient management in 11 additional patients (12.6%).
The Dutch cooperative randomized study of Herder et al.[47] randomly assigned 465 patients with suspected NSCLC to either traditional staging workup or a PET scan. PET was followed by histologic and or cytologic verification of lesions or further imaging and follow-up. The primary outcome was the number of tests and procedures to finalize staging and define operability. No statistically significant difference was found between the two groups for the mean number of tests to finalize staging. Secondary out-comes were the length of the diagnostic process, morbidity associated with staging procedures, and cost. The median time to diagnosis was statistically significantly shorter for the PET group (14 days versus 23 days, P<.0001). There was no difference in the morbidity associated with the staging procedures. The mean number of functional tests, noninvasive procedures, invasive procedures, and thoracotomies did not differ between the arms; however, the percentage of patients who needed more than one invasive test for lymph node staging and the numbers of mediastinoscopies were statistically significantly lower for the PET group. However, it is not clear from the report whether these outcomes were a priori or post hoc comparisons or whether statistical analyses were adjusted for multiple comparisons.
Results of Primary Studies of
18Fluorodeoxyglucose Positron Emission Tomography Accuracy in Non-Small Cell Lung Cancer StagingTwenty-two prospective observational studies,[34-46,51-59] 14 of which were included in ICES,[34-46,51] examined the use of PET in staging primary NSCLC ( Table 3 and Table 4 ). Most studies enrolled patients with potentially resectable NSCLC and used histopathologic results as the reference standard. The protocols for nodal sampling varied between the trials and were not always clearly described. The methods used for reporting PET scans as positive varied, with some studies visually interpreting the scan and others using semiquantitative methods, such as calculating the SUV. PET data were evaluated independently of the reference standard in 20 studies.[34-46,51,52,55-60] PET was included in the reference standard in one study,[51] which could have led to an overestimation of diagnostic accuracy. In four trials,[37,44-46] fewer than 100% of patients had confirmation of the diagnosis by the reference standard, which could have led to biased estimates of overall diagnostic accuracy. Eighteen of the studies[34-42,45,46,51,54-59] specified explicit criteria for defining a positive PET test result. Four studies[36-38,54] reported results using lymph nodes as the unit of analysis. The observations in these four studies are not statistically independent in that a patient with one positive lymph node is likely to have other positive lymph nodes, which may bias the estimates of diagnostic accuracy. Results from studies examining staging of the primary tumor were variable, as were the criteria used to determine a positive result (e.g., N0 versus N1/2/3 or N0/N1 versus N2/3). Sensitivity of PET for detecting distant metastases ranged from 82% to 90% and specificity ranged from 90% to 98%. Eight studies reported on the usefulness of PET for detecting unexpected distant metastases;[37,39-41,44-46,53] in these studies, PET detected distant metastases in 4%-17% of patients.
With the introduction of integrated PET-CT diagnostic imaging machines, additional observational studies have been published on the accuracy of this newer imaging technology. Cerfolio et al.[55] compared integrated PET-CT with dedicated PET for staging in 129 patients with NSCLC. Integrated PET-CT was more accurate than dedicated PET for predicting stage I and II disease and was a better predictor of overall tumor and node status; all differences were statistically significant. Integrated PET-CT was also more accurate overall for detecting N2 and N1 nodes, as well as defining T2, T3, N0, and N1 disease. Lardinois et al.[44] also compared integrated PET - CT with dedicated PET and found that integrated PET-CT improved the accuracy of tumor and node staging and the detection of metastases. In another comparison of integrated PET-CT with dedicated PET, Halpern et al.[57] found that integrated PET-CT was more accurate for assigning T stage and had greater accuracy for determining the overall tumor - node - metastasis stage. Shim et al.[58] compared integrated PET-CT with stand-alone CT and found that integrated PET-CT was statistically significantly more accurate than CT for nodal and overall staging but was not for tumor staging.
Finally, Oturai et al.[56] compared gamma-camera PET with dedicated PET and found no statistically significant difference for detecting primary pulmonary lesions or evaluating regional lymph nodes between the two techniques. Gamma-camera PET did have reduced sensitivity for detecting lymph nodes as compared with dedicated PET. Overall, the published literature indicates that PET-CT provides greater diagnostic accuracy in the staging of NSCLC than does PET alone.
Primary Non-Small Cell Lung Cancer: Accuracy of
Findings of Institute for Clinical and Evaluative Sciences
The ICES report concluded that evidence exists that PET is efficacious in predicting the histological status of mediastinal lymph nodes and in detecting pleural involvement and malignant pleural effusion for patients with carcinoma of the lung. The report also concluded that PET is more efficacious than CT in identifying abnormalities in mediastinal lymph nodes.
Results of Systematic Reviews
Fischer et al.[23] estimated the mean sensitivity and specificity independently for staging metastases in the mediastinum. The mean sensitivities and specificities were 0.83 (SE = 0.02) and 0.96 (SE = 0.01), respectively, for dedicated PET and 0.81 (SE = 0.04) and 0.95 (SE = 0.02), respectively, for gamma-camera PET. The authors concluded that PET is a valuable tool for staging NSCLC and noted that although its use for preoperative staging is strengthened by its high specificity, further examinations in populations with a lower prevalence of NSCLC are still required.
A meta-analysis by Birim et al.[21] included 17 studies[35,38-40,43,60-71] that compared PET with CT in detecting mediastinal lymph node metastases. The maximum joint sensitivity and specificity of PET from the SROC curve was 90% (95% CI = 86% to 95%). Birim et al.[21] concluded that PET was more accurate than CT imaging in detecting mediastinal lymph node metastases. The authors recommended that PET images be correlated with a CT scan because PET has limited ability to determine precise anatomic localization of mediastinal lymph nodes. A meta-analysis by Gould et al.[25] also concluded that PET is more accurate than CT (P <.001) for mediastinal staging in patients with potentially resectable NSCLC. For the 32 studies in the Gould et al.[25] metaanalysis in which the patient was the unit of analysis,[34,35,39-41,43,61-63,65,67-69,71-89] a maximum joint sensitivity and specificity of PET was calculated from the SROC curve as 86% (95% CI = 84% to 88%), which at a median specificity of 90% would correspond to a sensitivity of 81%. Gould et al.[25] also examined the use of PET for identifying mediastinal metastasis in patients with and without enlarged lymph nodes on CT in a metaanalysis of data from 14 studies[34,35,39,40,61,63,65,67,69,71,76,78,79,90] and found that PET was more sensitive but less specific when the CT scan showed enlarged mediastinal lymph nodes than when it did not. The authors concluded that biopsy should confirm positive PET findings before curative surgery is excluded as a treatment option and that negative PET findings should be interpreted in light of the patient's pretest probability of mediastinal metastases and whether CT reveals enlarged mediastinal nodes.[25]
Results of Primary Studies
Halter et al.[52] evaluated PET in staging mediastinal lymph nodes in 155 patients with pulmonary tumors. PET was associated with accuracies of 91%, 77%, 95%, and 100% for N0, N1, N2, and N3 disease, respectively. Verhagen et al.[53] assessed the reliability of PET for staging mediastinal lymph nodes in 66 patients with NSCLC. Although the negative predictive value for staging mediastinal lymph nodes was 71%, the negative predictive value for patients with positive N1 nodes and/or a centrally located primary tumor was only 17%, compared with 96% for patients with negative N1 nodes and a noncentrally located primary tumor.[53] Nomori et al.[54] measured the size of metastatic foci in lymph nodes for which the PET results were true positives and false negatives to determine the lower size limit of metastatic lymph nodes that PET can detect. Metastatic foci in the lymph nodes with true positive results had a mean size of 10 mm (range 4-18 mm), and those with false negative results had a mean size of 3 mm (range 0.5-9 mm). Lymph nodes with false positive results had a mean size of 12 mm (range 9-16 mm), and those with true positive results had mean size of 10 mm (range 6-15 mm). PET did not detect any metastatic foci smaller than 4 mm. Lardinois et al.[44] compared integrated PET-CT with dedicated PET and found that integrated PET-CT improved the accuracy of staging mediastinal metastases. Pozo-Rodriguez et al.[59] evaluated contrast-enhanced helical CT and PET, both alone and combined. Helical CT and PET performed similarly in the diagnostic accuracy of mediastinal staging, although the authors concluded that both tests are conditionally dependent and provide complementary information.[59]
Primary Non-Small Cell Lung Cancer: Accuracy of
Findings of Evidence-Based Reports
Although the ICES report did not address extrathoracic staging with PET, this topic was addressed in four other evidence-based reports. The Health Technology Board for Scotland report[22] evaluated 19 studies on the detection of distant metastases[34,37,39,40,51,60,63,69,84,87,91-99] and concluded that there is evidence that PET may be useful in staging patients believed to be free of distant metastases, especially in the detection of adrenal gland and bone metastases, but recommended that this be confirmed in controlled trials. In addition, a review by the National Institute for Clinical Excellence[26] provided an SROC curve for the detection of distant metastases and calculated a summary sensitivity of 93% and specificity of 96%. This review also found that an average of 15% of patients had unexpected distant metastases detected by PET.
Results of Primary Studies
Only one prospective study reported on the staging of extrathoracic metastases. Verhagen et al.[53] assessed the value of PET in detecting extrathoracic metastases in 72 patients with NSCLC. PET detected extrathoracic metastases in 15% (10/66) of patients in whom conventional staging showed no evidence of metastases.
Staging of Small-Cell Lung Cancer with
Results of Primary Studies
Three prospective studies[14-16] examined the use of PET in staging primary SCLC ( Table 6 ). The reference standards varied among the studies, and none of the studies confirmed all lesions with histologic results. Brink et al.[15] confirmed only 20% of lesions with histopathologic results. PET was evaluated independently of the reference standard in two of the studies.[14,15] Only one study clearly specified explicit criteria for defining a positive PET test result.[15] Sensitivity for staging extensiveversus limited-stage disease ranged from 89% to 100% and specificity ranged from 78% to 95%. Chin et al.[16] compared PET with conventional staging and reported that PET agreed with conventional staging in 15 of 18 cases (83%). PET showed more extensive disease in two of the three patients for which PET and conventional staging disagreed. These data suggest that total-body PET may be useful in the staging of SCLC.
Integrating
For patients with an established diagnosis of lung cancer (NSCLC or SCLC), accurate staging is essential to determine appropriate treatment decisions. Conventional staging procedures are imperfect in their ability to define those patients with the potential for cure with surgery or locoregional chemoradiotherapy approaches. PET has the potential to improve staging accuracy, but health technology assessment reports have concluded that the amount of improvement in diagnostic accuracy of PET in staging NSCLC is difficult to define due to the variations in study quality and the lack of direct evidence on whether PET improves patient outcomes.[22,26] Meta-analyses found sensitivity to range from 81% to 90% and specificity to range from 89% to 90% for the distinction between N0-1 and N2-3 patients.[21,25,100] Accuracy studies have had similar results, with PET results being superior to CT imaging for mediastinal staging. Studies that interpret PET images with CT results have higher accuracy than when PET is interpreted independently.[40,44] However, results from Nomori et al.[54] indicate that PET is unable to detect metastatic foci smaller than 4 mm. On the other hand, false positives with respect to staging the mediastinum can also occur with infection and inflammation. The available evidence indicates that a positive PET scan of the mediastinum must be confirmed histologically or cytologically to ensure that patients are not denied potentially curative surgery. It should also be recognized that false negative results can occur when the primary tumor obscures mediastinal lymph nodes or the metastatic foci are small.
PET has been found to have high accuracy (89%-96%) for detecting distant metastases and has detected extrathoracic metastases in patients in whom conventional imaging showed no evidence of distant metastases.[99] The role of PET in the evaluation of distant metastases appears to be greatest for adrenal and bone metastases. PET is not useful for detection of brain metastases due to the high glucose uptake of normal brain tissue.
A number of factors could contribute to the apparent discrepancy between two of the trials evaluating the value of preoperative PET assessment. One factor is the difference in the patient populations included in the trials. The PLUS trial[48] entered patients with both suspected and proven NSCLC based on clinical and not pathologic assessment. As a result, this study included patients with both benign and malignant lesions, whereas the Australian trial[49] only included patients with histologically or cytologically proven NSCLC prior to randomization. Moreover, 29% of patients in the PLUS trial had clinical stage III disease on entry to the study, whereas the Australian trial only included patients demonstrating clinical stage I or II disease. The approach to the management of patients with early-stage lung cancer also differed between the two studies. Patients in the Australian trial with stage IIIA disease underwent surgery, whereas thoracotomy was considered to be futile in the PLUS trial if the patients had stage IIIA/N2 disease. Finally, the definition of a futile thoracotomy (benign disease, exploratory thoracotomy only, pathological stage IIIA (mediastinal node positive) or IIIB disease or postoperative relapse or death within 12 months of randomization) in the PLUS study differed from the Australian trial definition of avoided thoracotomies (patients who were able to avoid thoracotomy as determined at the discretion of the surgeon). As a result, the different designs of these studies and their contradicting results contribute to the confusion concerning the clinical utility of PET in the management of patients with early-stage lung cancer.
PET has not been studied as extensively in staging patients with SCLC. PET appears to have good accuracy (83%-99%) in differentiating extensiveversus limited-stage SCLC,[14-16] which could help in deciding which patients would be the best candidates for combined modality (chemoradiotherapy) therapy in limited-stage disease.
Evaluation of new imaging techniques is important because high costs, an increasing demand for healthcare, and limited budgets have resulted in a need to prioritize the implementation of new techniques.[101] PET scanning could improve the results of surgical therapy for early-stage lung cancer by excluding patients from surgical resection who have evidence of metastatic disease that is not evident by standard preoperative staging procedures. Similarly, the results for the management of locally advanced disease might also be expected to improve because of the addition of patients with minimal contralateral nodal disease that precluded surgery. Moreover, if PET imaging spares patients from the potential morbidity and risk of mortality from an unnecessary surgical procedure or chemoradiotherapeutic intervention, it would not only have a substantial impact on individual patients but allow for more efficient and effective utilization of limited health care resources.
One of the limitations of this systematic review of PET in lung cancer is that the conclusions that can be drawn are constrained by the available evidence, and the available evidence may not be directly relevant to current practice. For example, the type of PET device used in the published trials impacts on the generalizability of the findings from the review. Some studies reported using gamma cameras or coincidence imaging devices, which are rapidly being replaced by higher precision scanners (e.g., PET/CT hybrid). Although coincidence scanners are no longer marketed, these devices continue to be used in some settings. We do not consider this to be a serious limitation of this review as many of the trials included in this review have been published since 2000 and have evaluated devices that are in standard use today.
Future research is needed to determine not only if PET should be integrated into the standard staging and diagnostic process of lung cancer but also how PET would be incorporated into the staging algorithm. The Ontario Clinical Oncology Group is currently conducting two prospective randomized controlled trials on the use of PET, and the province of Ontario has established a registry study of PET in patients with SPN.
Fine needle aspiration biopsy should continue to be the firstline diagnostic approach in the workup of SPN because of its ability to provide a definitive diagnosis in a high percentage of patients and its relative safety. However, based on this review, PET can be considered to be useful in those situations in which a biopsy is inconclusive or contraindicated. For potential surgical candidates, it is essential that PET-positive mediastinal lesions be confirmed histologically or cytologically by mediastinoscopy or other diagnostic procedure to verify that PET-positive mediastinal lesions are due to cancer. Tissue confirmation is essential to ensure that patients are not denied potentially curative surgery. Solitary extrathoracic sites of PET positivity should also be confirmed to be metastatic if at all possible to ensure that patients are not inappropriately denied the chance of potentially curative therapy.
Source : http://www.medscape.com/viewarticle/573893_
No comments:
Post a Comment