| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Ann Thorac Surg 2007;84:959-966
© 2007 The Society of Thoracic Surgeons
a Department of Thoracic Surgery, European Institute of Oncology, Milan
b Department of Radiology, European Institute of Oncology, Milan
c Department of Nuclear Medicine, European Institute of Oncology, Milan
d Department of Epidemiology and Biostatistics, European Institute of Oncology, Milan
e Division of Pathology, European Institute of Oncology, Milan
f Division of Scientific Direction, European Institute of Oncology, Milan
g School of Medicine, University of Milan, Milan, Italy
Accepted for publication April 16, 2007.
* Address correspondence to Dr Veronesi, Division of Thoracic Surgery, European Institute of Oncology, Via Ripamonti 435, Milan, I-20141, Italy (Email: giulia.veronesi{at}ieo.it).
Presented at the Forty-third Annual Meeting of The Society of Thoracic Surgeons, San Diego, CA, Jan 29–31, 2007.
 |
Abstract |
|---|
 Top Abstract IntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Methods: A total of 5200 asymptomatic current or former smokers (
20 pack-years) older than 50 years of age were enrolled in a single-institution screening trial using annual LD-CT. Growing nodules and those with a maximum diameter exceeding 8 mm were studied with CT-PET. Transthoracic needle biopsy was not a routine part of the protocol.
Results: During the first year of study, 157 subjects underwent CT-PET, 66 of whom underwent surgical biopsy. Of the 58 lung cancers found on surgical biopsy, 51 were positive (standard uptake value >2.0) and seven were negative for malignancy by CT-PET. Sensitivity was 88% overall, but 100% in the subgroup with solid nodules of 10 mm or more. Among the 8 patients with benign disease at surgical biopsy, CT-PET was positive in 6 and negative in 2.
Conclusions: CT-PET is a highly promising modality for identifying potentially malignant lesions in screening-detected lung nodules and appears particularly useful as an alternative, in the screening setting, to invasive procedures for the further investigation of uncertain nodules. Our findings also indicate that the standard uptake value threshold for positivity should be lowered for small nodules (<10 mm). Longer follow-up and larger prospective studies are necessary to confirm these preliminary findings.
|
Introduction |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
Noncalcified nodules are detected by low-dose CT in 40% to 70% of screened subjects [6, 8, 9]. It has recently been shown that micronodules (<5 mm in diameter) can be safely checked once a year [10, 11], thereby considerably reducing the recall rate and screening costs. However, nodules >5 mm, which require further investigation, occur in 10% to 15% of screened subjects [7, 8, 12], and there is no agreement on the best way of managing them.
In most screening studies, percutaneous fine-needle aspiration is used to investigate suggestive nodules [5]. This technique requires considerable expertise, however, particularly to biopsy small deep lesions, and is associated with low sensitivity for benign lesions [13, 14] and nonnegligible morbidity [15].
Positron emission tomography (PET) with F-18 fluoro-2-deoxyglucose (FDG) is a noninvasive technique used for identifying malignant lung lesions [16] with sensitivity of 80% to 100% in many series [17–21]. PET, however, has poor sensitivity for small lesions and those with ground glass opacity (GGO), which are often bronchioloalveolar carcinoma, on CT. Nomori and colleagues [22] found that sensitivity and specificity for GGO lesions were only 10% and 20%, respectively, and for malignant nodules smaller than 1 cm, PET had a sensitivity of 0%.
Integrated PET and CT (CT-PET) is increasingly used to image and characterize lesions in cancer patients because of its improved instrument resolution [23]. The utility of CT-PET in a screening setting has not been extensively investigated, however [9, 24]. We conducted a prospective study to assess the sensitivity, specificity, and accuracy of CT-PET in the differential diagnosis of lung nodules detected by low-dose spiral CT in high-risk individuals at baseline in the Continuing Observation of Smoking Subjects (COSMOS) lung cancer screening trial [12].
|
Patients and Methods |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
20 pack-years) or had stopped smoking in the 10 years before study entry and did not have any malignant disease except adequately treated nonmelanoma skin cancer. The protocol required once yearly low-dose CT scans for 5 years, baseline spirometry, compilation of two self-reporting questionnaires investigating respiratory symptoms (British Medical Research Council questionnaire, 1960) [25] and smoking habits, and blood sampling to perform genetic marker and proteomic studies in relation to lung cancer risk. The study was approved by the Ethics Committee of European Institute of Oncology. All recruited volunteers gave written consent to annual CT for 5 consecutive years and agreed they would not be informed of detected lesions with a maximum diameter of 5 mm or less, although such lesions would be specifically investigated on the next scan. Nodules 5 mm or less were scheduled for repeat CT a year later. Nodules between 5.1 and 8 mm were scheduled for repeat CT 3 months later. Noncalcified nodules of maximum size exceeding 8.1 mm (or growing lesions <8 mm after repeat scan) were scheduled for PET-CT unless they appeared benign, as will be described in subsequent sections.
Small lesions suggestive of malignancy (growing or CT-PET–positive) were scheduled for surgical biopsy and additional interventions. In addition to PET scan, nonresectable lesions (advanced stage) received CT with contrast of the upper abdomen, chest, and brain, and further interventions as determined by imaging outcomes.
Lesions of any size considered to be from infection were treated with oral antibiotics for 10 days and CT was repeated a month later. Other lesions were defined as benign-based on morphologic CT features [26]—oval shape (maximum axial diameter greater than twice minimum diameter), thickening of fissures (liquid density, inside apical scar)—and were scheduled for repeat low dose CT 3 months later.
Low-Dose Computed Tomography
The CT equipment was a High Speed Advantage (General Electric, Milwaukee, WI) 8-slice or 16-slice multidetector scanner. No contrast was used for screening scans, which were taken in a single breath (machine setting: 140 kVp, 30 mA, 1.75:1 pitch ratio, and 2.5-mm slice thickness). Images were retro-reconstructed using standard and lung algorithms at 2.5-mm intervals. The dose equivalent per patient was estimated at 0.81 mSv. A radiologist read the images on a workstation (Advantage Windows 4.2, General Electric), using lung parenchyma windows with maximum intensity projection reconstruction and mediastinum windows. Findings of noncalcified lung nodules were discussed at meetings with radiologists, thoracic surgeons, and nuclear medicine physicians. Nodules were classified as solid, partially solid, or nonsolid [27].
Computed Tomography-Positron Emission Tomography
Patients were fasted 6 hours. After their blood glucose was confirmed to be less than 150 mg/dL, they were administered 5 MBq/kg of FDG intravenously. They then waited in calm conditions (minimum movement, no speaking) for 50 to 60 minutes. Images were acquired with a combined CT-PET in-line system (Discovery LS, GE Medical Systems) consisting of an Advance NXi PET scanner and an 8-slice Light Speed Plus CT scanner.
Patients were positioned headfirst supine and moved to the CT scanning position. A scout scan was acquired to define the axial imaging range, which for whole-body CT-PET typically extended from the lower jaw to the upper thighs. CT settings were 140 kVp and 80 mA. Patients were instructed to breathe normally. The CT image was matched to the section thickness of the PET image, which covered the same axial field as the CT.
After completion of CT, the examination table was advanced by 60 cm into the PET gantry (axes of CT and PET gantries in line). PET acquisition time was 4 minutes per table position, with a complete scan time of about 30 minutes. PET image data sets were reconstructed iteratively, with segmented correction for attenuation using the CT data [28].
The CT-PET scan results were considered positive if the maximum standard uptake value (max-SUV) of FDG exceeded 2.0 in a region of interest (ROI) automatically calculated on lesions identified by CT. We used this low threshold because the consequences of a false-negative result (treatment delay) were more undesirable than the consequences of a false-positive result (unnecessary biopsy or surgery). We also assessed the diagnostic accuracy of other thresholds.
Surgical Biopsy and Treatment
Patients with a suggestive lesion underwent diagnostic wedge resection. A video thoracoscopic approach was used whenever lesion site and size allowed, otherwise the approach was through a standard muscle-sparing thoracotomy. The wedge underwent intraoperative frozen section examination. If the pathology result was positive, anatomic resection with curative intent was performed immediately, usually associated with radical lymph node dissection.
Statistical Analysis
The CT-PET results were compared with histopathologic findings on surgical biopsy and with at least 12 months of clinical follow-up in the other cases. SUV distributions in patient subgroups are presented using box and whisker plots. The differences in median SUV between subgroups were assessed using the nonparametric two-samples median test. The diagnostic value of CT-PET was assessed by receiver operating characteristics (ROC) curves of max-SUV. Nonparametric ROC curves were generated by plotting sensitivity versus one-specificity, so that an ideal test has sensitivity and specificity of one. The area under the ROC curve (AUROC) was used as a measure of the diagnostic efficiency of a test. Sensitivity, specificity, positive and negative predictive values, and accuracy are presented for several cutoff values of SUV. The analyses were performed with SAS software (SAS Institute, Cary, NC). All p values are two-sided.
|
Results |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
At the baseline screening, 2197 volunteers had one or more noncalcified nodules of 5 mm or less and were scheduled for annual follow-up, and 354 had a nodule between 5 and 8 mm and were scheduled for low-dose CT 3 months later, and 9 subsequently received CT-PET for a growing nodule. Another 205 volunteers had a nodule exceeding 8 mm at baseline, 148 of whom received CT-PET and 65 did not. Those who did not receive CT-PET had a lesion compatible with infection and were prescribed antibiotics and underwent repeat low-dose CT 3 months later, or had a lesion with clearly benign morphologic features and were scheduled for repeat low-dose CT.
Thus, 157 patients underwent CT-PET. Their nodules were solid in 121 cases, partially solid in 30, and not solid in 6. A total of 58 lung cancers were diagnosed: 55 after baseline screening and three others at screening 1 year later. The mean size was 20.5 mm (range, 6 to 60 mm). There were 41 (71%) adenocarcinomas, eight (14%) squamous cell carcinomas, three (5%) typical carcinoids, three (5%) small cell lung cancers, and three (5%) not otherwise specified non-small cell lung cancers. Three patients with adenocarcinoma had multifocal disease. Forty-one (69%) subjects were stage I; 3 (5.2%) were stage II, 11 (19%) were stage III, and 3 (5.2%) were stage IV.
The max-SUV was available for 154 of 157 examinations; in the remaining three, FDG-PET was performed elsewhere and the SUV was not available, although all three were reported suggestive of cancer. Table 1 summarizes the variables of diagnostic accuracy for the CT-PET examinations, with max-SUV exceeding 2.0 considered a positive result. Of the 58 lung cancers found on surgical biopsy, 51 were positive and seven were negative on CT-PET. Overall sensitivity, specificity, and accuracy of the examination were 88%, 93%, and 91%, respectively. Table 1 also shows results according to nodule size and type. For solid nodules, sensitivity, specificity, and accuracy of CT-PET were 95%, 94%, and 94%, respectively; the corresponding values for nodules of 10 mm or more were 89%, 91%, and 90%. CT-PET performance was best for solid nodules of 10 mm or more (sensitivity 100%, specificity 90%).
|
|
|
|
|
|
|
|
|
Comment |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
For nodules of less than 10 mm, our diagnostic variables were somewhat better (83% sensitivity, 100% specificity, and 95% accuracy) than overall, and in apparent contrast with Nomori and colleagues’ [22] finding that all lesions up to 1 cm were negative on PET, regardless of histology. Likely explanations are improved sensitivity of CT-PET compared with PET alone and a possibly higher proportion of difficult-to-diagnose GGO nodules in their series. In our series, 10 of 12 solid lesions smaller than 10 mm were positive on CT-PET; the remaining 2 were false negatives with max-SUVs of 1.6 and 1.7, higher than SUVs of true negative lesions.
From our analysis of the prognostic values of different max SUV thresholds, we found that adopting a max-SUV of 1.5 for lesions smaller than 10 mm increased sensitivity without decreasing specificity. We therefore suggest using 1.5 as the max-SUV cutoff for small lesions. By contrast, for solid nodules of 10 mm or more, sensitivity was 100% and specificity was 90% when a max-SUV of 2.0 was used as the cutoff.
The 2005 study of Port and colleagues [32] found that PET had a sensitivity of only 45% for the diagnosis of patients with lung cancers of 2 cm or less and was of "no demonstrable benefit." It is important to note, however, that Port and colleagues used a higher SUV cutoff than we did (2.5 versus 2.0), and sensitivity dropped to 73% when we adopted a 2.5 SUV cutoff. Furthermore, their mean tumor size was smaller than the mean tumor size in our study (14 mm versus 20.5 mm), and this may have also have contributed to sensitivity reduction.
We found that nonsolid nodules do not have altered FDG metabolism, so PET is nondiagnostic; however less than 10% of malignant lesions in our study were nonsolid. It is also noteworthy that most of nonsolid nodules assessed by CT-PET and biopsied surgically were malignant. The implication is that the radiologic characteristics of nonsolid nodules suggestive for bronchioalveolar adenocarcinoma are so typical that CT alone is enough to diagnose malignancy. Thus, no diagnosis delays occurred in nonsolid nodules due to false negative CT-PET findings. We also found that mixed nodules with a solid component (partially solid) had CT-PET characteristics similar to solid nodules, and the examination had high sensitivity, specificity, and accuracy for these nodules.
Comparison of false-positive and true-positive nodules revealed they differed significantly in terms of median max-SUV, and that false-positive (benign) lesions never had a max-SUV above 3.01, even if the lesion was large. This finding may be useful for interpreting PET results, and perhaps justifying postponement of surgical biopsy until a subsequent follow-up CT suggests that it is a necessity.
Collateral findings included a significant direct association between nodule size and SUV, and that SUV was significantly higher in squamous cell carcinoma than in all other tumor types combined.
It is important to assess CT-PET in relation to other diagnostic modalities in the screening setting. Bronchoscopy with brush cytology must be excluded because of its low diagnostic yield for lesions smaller than 2 cm [33], whereas transthoracic needle aspiration biopsy is characterized by false-negative levels of up to 30% [13] and a complication rate of up to 30% in some series [14]. Nodule-enhancement CT has also been investigated [34], but the authors concluded that FDG-PET was preferable to nodule-enhancement CT in evaluating indeterminate lung nodules because of its much higher specificity and only slightly reduced sensitivity.
In the future, it is likely that we will be able to use serum or sputum biomarkers to define the risk of lung malignancy in individuals exposed to carcinogens, enabling the timing of follow-up CT scans to be optimized and also suggesting the best type of invasive investigations. In the meantime, CT-PET appears to be the most promising noninvasive imaging procedure in the screening setting because of its high specificity and sensitivity as well as its ability to reduce the number of invasive procedures performed and their attendant complications.
The important additional dimension of the cost-effectiveness of screening CT-PET has also been addressed. The study of Lejeune and colleagues [35] used a decision analysis model to compare costs in three management scenarios for solitary lung nodules: wait and watch, PET with anatomic CT, and CT plus PET. They found that CT plus PET was not only the most effective strategy of the three but also had the lowest incremental cost-effectiveness ratio. Gugiatti and colleagues [36] compared the addition of PET with traditional workup (CT, transthoracic needle aspiration biopsy, or surgical biopsy) for solitary lung nodules. They found addition of PET was associated with a cost reduction of about 
50 per patient because it reduced inappropriate invasive diagnostic investigations and their complications.
In the present study, we chose to use only instrumental estimates of FDG uptake as an indicator of nodule nature and did not include operator (visual) estimates. This choice is appropriate in view of a hypothesized extension of the study to a regional or national screening program for lung cancer.
A limitation of CT-PET is that it may not easily be applicable to populations where histoplasmosis is endemic, because the false-positive rate would be high. A limitation of our study is the short 1-year follow-up, and it is possible that some very slow growing cancers tumors were not recognized.
The COSMOS study is ongoing, and second-year data are being analyzed. We are expanding our assessment of CT-PET in this setting by investigating relations between CT-PET findings and the growth rate of malignant lesions, evaluating the staging role of the technique (implications for minimally invasive surgery, resection extension and lymph node dissection), comparing CT enhancement and CT-PET findings, further evaluating the utility of SUV, and also performing a cost-effectiveness analysis.
CT-PET is characterized by high sensitivity, specificity, and accuracy in the diagnosis of malignant lung lesions detected by low-dose CT in the screening setting. It is particularly promising, because it reduces the use of invasive diagnostic procedures. Our findings also indicate that the SUV threshold for positivity should be lowered for small nodules (less than 10 mm).
|
Discussion |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
DR VERONESI: That is a good question. In fact there was a little bit of discussion with the region that had to pay for the CT-PET. We had grants for the CT screening but not for the diagnostic work-up of screening-detected nodules that in our opinion should have been paid by the national health system. In our experience, we reached a compromise: we paid half of the CT-PET and the other half was paid by the national health system.
DR GEORGE B. HAASLER (Milwaukee, WI): In the nodules that were PET-positive, how many of the mediastinal lymph nodes were positive in that setting? In other words, what was the sensitivity for the study for those lymph nodes for the nodules that were positive? Also, did the degree or the likelihood of lymph node positivity in the mediastinum relate at all to how avid the nodules were?
DR VERONESI: The staging accuracy of PET was not the main objective of this paper, so I have only preliminary data and details will be analyzed in a near future. Anyway we had a high sensitivity for nodal staging in the stage III disease that were about 15% of the tumors found with screening. The cases that were positive at PET scan underwent biopsy with mediastinoscopy and induction chemotherapy.
DR FREDERIC W. GRANNIS (Duarte, CA): Despite the recent results of the ELCAP study, which showed excellent results in long-term survival, we’re having a hard time convincing CMS, the American Cancer Society, et cetera, in the United States that lung cancer screening is safe and effective, and I would congratulate you on this study that appears to show what we have emphasized, that lung cancer screening is a process, not just a test, that we must consider the false positives within the context of the protocol of diagnosis, and, finally, that good protocols prevent unnecessary operations. How many patients in your series had a surgical operation for a nonmalignant pulmonary nodule out of your 5000 screened patients?
DR VERONESI: I presented these data at the ASCO meeting last June; the final result of our first year of screening concluded with 13% of surgical procedures for benign disease. We had seven cases of benign disease detected at surgery with minor resection out of 62 surgical procedures. We compared these data with our standard practice in our division and found that it was a little bit lower than our standard result in a nonscreened population.
DR ARA VAPORCIYAN (Houston, TX): I enjoyed your presentation very much. I wanted you to comment on the generalizability of your findings to areas of the world that have more endemic infectious diseases, such as histoplasmosis and other diseases, which really increase the rate of false positivity on a PET.
DR VERONESI: I think that the implication of this problem that you have underlined could result in a higher percentage of invasive procedures for benign disease; however, the optimization of the diagnostic protocol with repeated low-dose CTs in positive cases to see the development of the lesions associated to the use of antibiotics before one CT scan and the other will limit the false positives.
DR RICARDO S. SANTOS (Pittsburgh, PA): If we all do lung CT screening programs, we’re going to find smaller and smaller lesions, and then we present the case with a very small lesion that you do a lobectomy with radical lymphadenectomy. I know that this is a little bit out of the scope, but I would like to hear your comment about which type of operation you would recommend to patients in whom their lesions are less than 2 cm. I have a lot of data from the literature, the Japanese literature and also the American literature, that these patients probably would not need a lobectomy, that segmentectomy could be enough. Could you please comment on that? Congratulations on your presentation.
DR VERONESI: Thank you. My personal opinion is that for a lesion smaller than 15 mm, in a peripheral position, and with a low level of SUV, probably a wide wedge resection would be an appropriate treatment, but this should be confirmed and studied in randomized trials. For the moment, I think that as this screening process is in an experimental phase, we must use a standardized technique for treatment not to add other variables. For this reason, we still consider lobectomy with radical lymph node dissection the standard extension of resection for these cases.
DR NASSER K. ALTORKI (New York, NY): Your results are obviously spectacular. I may have missed this, but did you draw a distinction between findings on the baseline and the repeat screen?
DR VERONESI: I didn’t bring those data in this setting, but we are analyzing the results of the annual screening. My feeling is that small lesions that have a very slow growth rate are more often negative at PET scan. So maybe the data of the annual screening may produce a lower level of sensitivity because a certain number of nodules have grown very, very slowly.
DR ALTORKI: It would also improve your specificity compared to the baseline.
DR VERONESI: Probably yes.
|
References |
|---|
|
TopAbstractIntroductionPatients and MethodsResultsCommentDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. De Cicco, M. Bellomi, M. Bartolomei, G. Carbone, G. Pelosi, G. Veronesi, T. De Pas, L. Spaggiari, and G. Paganelli Imaging of Lung Hamartomas by Multidetector Computed Tomography and Positron Emission Tomography. Ann. Thorac. Surg., December 1, 2008; 86(6): 1769 - 1772. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Veronesi, M. Bellomi, P. Scanagatta, L. Preda, C. Rampinelli, J. Guarize, G. Pelosi, P. Maisonneuve, F. Leo, P. Solli, et al. Difficulties encountered managing nodules detected during a computed tomography lung cancer screening program J. Thorac. Cardiovasc. Surg., September 1, 2008; 136(3): 611 - 617. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R Carbone, E Balleari, and R Ghio Accuracy in suspicious lung infiltrations Thorax, June 1, 2008; 63(6): 568 - 568. [Full Text] [PDF]
|
|
|
|
|
|
T Targowski and K Jahnz-Rozyk Authors' reply Thorax, June 1, 2008; 63(6): 568 - 568. [Full Text] [PDF]
|
|
|
|
 |
|
 P. E. Van Schil, J. M. Hendriks, and P. Lauwers Editorial comment: Staging of ipsilateral pulmonary nodules: back to the future Eur. J. Cardiothorac. Surg., March 1, 2008; 33(3): 485 - 486. [Full Text] [PDF]
|
|
|
|
 |
|
 G. Giaccone 18Fluorodeoxyglucose Positron Emission Tomography, a Standard Diagnostic Tool in Lung Cancer J Natl Cancer Inst, December 5, 2007; 99(23): 1741 - 1743. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
= outliers.)
