Accuracy of Helical Computed Tomography and [18F] Fluorodeoxyglucose Positron Emission Tomography for Identifying Lymph Node Mediastinal Met
http://www.100md.com
《临床肿瘤学》
the Clinical Epidemiology Unit, Department of Pulmonology, Department of Thoracic Surgery, Department of Radiology, Hospital Universitario 12 de Octubre, and the Centro PET Complutense, Madrid, Spain
ABSTRACT
PURPOSE: Computed tomography (CT) and [18F] Fluorodeoxyglucose positron emission tomography (FDG-PET) are considered suitable methods for the noninvasive staging of the mediastinum. Our study was intended to estimate the efficacy of contrast-enhanced helical CT (hCT) and FDG-PET, alone and combined, in the diagnosis of lymph node mediastinal metastases.
METHODS: This study was a prospective and blind comparison of the efficacy of hCT and FDG-PET with two alternative reference standards, mediastinoscopy, and mediastinoscopy plus thoracotomy plus a 6-month follow-up to diagnose lymph node mediastinal metastases in 132 consecutive patients with potentially resectable non–small-cell lung cancer (NSCLC). The metastatic disease was assessed histopathologically. Further clinical information was obtained postoperatively after a median follow-up of 42 months.
RESULTS: The prevalence of cN2,3 is 0.28. For hCT the sensitivity and specificity are 0.86 (95% CI, 0.70 to 0.93) and 0.67 (95% CI, 0.56 to 0.75), for PET 0.94 (95% CI, 0.81 to 0.98) and 0.59 (95% CI, 0.49 to 0.68), and for hCT and PET combined in-parallel 0.97 (95% CI, 0.84 to 0.99) and 0.44 (95% CI, 0.34 to 0.53), which translate into a negative predicted probability of 0.98 (95% CI, 0.88 to 1.00). The crude diagnostic odds ratio of PET in the total sample studied is 13.1, in the subgroup hCT+ 11.04 (3.0 to 40 0.1), and in the hCT- 3.5 (0.5 to 21.5). Similar results were obtained for hCT stratified by PET.
CONCLUSION: hCT and PET perform similarly in the mediastinal staging of NSCLC, both tests are conditionally dependent and provide complementary information, and their diagnostic value mainly resides on the negative results.
INTRODUCTION
Lung cancer is the most common form of cancer.1 In the year 2000 the crude incidence rate in Spain was 86.7/105 in men and 7.7/105 in women. Lung cancer is a lethal disease with a 5-year survival rate of 15% overall and around 40% for those operated, that varies with the extent of the disease.2 In the year 2000 the Spanish National Center of Epidemiology reported 15,432 and 1,876 male and female deaths from lung cancer, an age adjusted mortality rate of 48.23/105 and 4.85/105, and a lifetime risk of dying of 5.56% and 0.51%, respectively.3
Surgical resection is the accepted standard of care for patients with non–small-cell lung cancer (NSCLC), but only approximately 25% are eligible for surgery.4 Once the surgical candidates are selected mediastinal staging is mandatory because up to 50% still have regional metastasis.5 Staging involves a considerable effort in terms of number of tests, duration, and cost.
The standard for invasive staging is mediastinoscopy/tomy (Ms-t).6 Computed tomography (CT) came up in the late 1970s as a noninvasive suitable staging diagnostic technique, but the papers published since then have shown conflicting results, and a meta-analysis published in 1990 concluded that no additional significant advances were to be expected from CT.7 In the early 1990s, positron emission tomography with [18F] Fluorodeoxyglucose (FDG-PET) emerged as a promising staging method because of its ability to detect areas of high FDG metabolic activity consistent with malignancy. The published reports generally conclude that PET is superior to CT for the mediastinum and point out the potential of PET to detect unexpected distant metastases.8 They rarely inform of how CT and PET perform together.
The goal of the present study is to determine the diagnostic efficacy of the helical-enhanced CT of the chest and upper abdomen (hCT) and the whole–body FDG–PET, used alone and combined in parallel, in the mediastinal staging of potentially operable NSCLC patients.
METHODS
Study Design
This is an observational study, cross-sectional with concurrently gathered information, performed between February 1999 and May 2001 at the Hospital Universitario 12 de Octubre (HU12X), a tertiary level hospital that provides health care for a population of 750,000 in Madrid. The target condition is metastatic disease in mediastinal lymph nodes, the index tests are hCT and PET, alone and combined in parallel; there are two reference standards: mediastinoscopy (Single Reference Standard-SRS) and mediastinoscopy plus thoracotomy plus a 6-month follow-up (Composite Reference Standard-CRS).
The study was designed to mimic the standard of clinical practice at the HU12X, having hCT and PET scans performed concurrently and apart from the standard clinical work-up and keeping the clinical decision makers unaware of their results. Both tests had to be done within 72 hours before the scheduled mediastinoscopy, and Ms-t had to be performed by the same two surgeons; otherwise, the patient did not enter the study for reasons that will be denominated as logistical. Radiologists and nuclear medicine physicians were blinded to clinical information and to PET and hCT scans, respectively. Two independent readings were obtained from each test and discrepancies solved by consensus. Following the study design, all clinical decisions were made after the patient was given the current standard of clinical practice and staged accordingly; PET or hCT images that suggested systemic metastases neither were notified to clinicians nor had to be histologically confirmed. The study was approved by the Ethics Committee of the HU12X, and the study subjects were presented with a form for written informed consent.
Participants
The eligible subjects were patients aged over 17, with histologically diagnosed NSCLC, staged I, II, and selected III, and potentially resectable. The exclusion criteria were allergy to iodine contrast and hyperglycemia over 9 mmol/L on the day PET was performed. The study subjects were all the consecutive eligible patients referred for Ms-t during the study period, who did not meet any of the exclusion criteria and gave their written informed consent.
Study Procedures
According to the standard of clinical practice of the HU12X, all patients histologically diagnosed from NSCLC underwent a study of operability consisting of a complete medical history and physical examination, routine hematologic and biochemical analyses, chest x-ray, ECG, bronchoscopy, conventional CT of the chest and upper abdomen, and lung function tests. Pertinent tests were performed whenever distant metastases were suspected. All potentially operable patients were subjected to cervical mediastinoscopy with parasternal mediastinotomy, if the tumor settled on the left upper lobe or left main bronchus. The Ms-t negative cases were referred to thoracotomy for pulmonary resection and mediastinal sampling.9 Staging was done according to the revised 1997 International System for Staging Lung Cancer.10
In addition to and apart from the standard work-up, the study patients undertook both an hCT and a whole-body PET scan. The hCT was performed with a 120 to 140 KV, 200 to 250 mA Toshiba X-Press spiral scan (Toshiba Medical Systems, Tokyo, Japan). Sections 7 mm thick from supraclavicular to infrarenal regions were obtained with a pitch 1,4 and a reconstruction interval of 5 mm. Each patient was administered intravenously 100 cm3 of a 0.117 mmol/cm3 iodine contrast solution (Opitray 300, Tyco Healthcare) at a rate of 2 cm3/s. There was a time lag of 30 seconds between the start of the perfusion and commencement of the scanning. The information gathered was the number and size of mediastinal lymph nodes detected. An hCT was considered positive (hCTN2,3) if the nodal shortest diameter was 10 mm or larger, and negative otherwise (hCTN0,1).
PET was performed on all patients at the Centro PET Complutense (Madrid), a private facility and the first PET center installed in Spain (1995). The first 47 scans were done with a POSICAM HZL-R camera (Positron Corporation, Houston, TX) with a field of view of 15 cm, a full width at half-maximal resolution of 5 mm, and correction for attenuation in thorax. The remaining 85 patients were scanned with an ECAT EXACT 47 camera (Siemens-CTI, Erlangen, Germany) with whole body attenuation, a field of view of 16.2 cm, a spatial resolution of 4.5 (2D), and 4.6 mm (3D). 18FDG was synthesized at the PET Center in a 12 Mev cyclotron (Oxford Instruments, Oxford, United Kingdom), following Hamacher's method.11 Examinations were performed after 6 hours of fasting. The transmission study (slices of width 16.6 cm) was done with a germanium-68 source during 20 minutes. Then, the patient was injected an intravenous dose of 370 MBq 18FDG. The emission study commenced 45 minutes after the injection and lasted for 20 minutes. Slices of width 16.2 cm were subsequently obtained with a 5 minutes exposition each, covering from the first cervical vertebra to the groins and corrected for attenuation with a Sun Ultra 60 computer 2 x 450 Mhz with an iterative reconstruction type weighted attenuation OSEM. Results were considered positive (PETN,2,3) when at least one definite, localized area of higher uptake than the mediastinal background was present, and negative otherwise (PETN0,1).
The result of the hCT and PET combined (in parallel) test was considered positive if any one of the components was positive and negative if they were both negative.
All mediastinoscopies were performed by the same two surgeons, who did a systematic sampling of normally and abnormally appearing mediastinal lymph nodes and fat. Stations 2, 4-superior and 4-inferior (left and right), and 7 were explored via cervical mediastinoscopy, and stations 5 and 6 via mediastinotomy. Thoracotomies were performed by seven thoracic surgeons. The samples obtained were stained with hematoxylin-eosin and examined histopathologically. Results were positive whenever metastases to lymph nodes were present.
Statistical Analysis
A minimum of 36 positive Ms-ts were needed to detect a sensitivity and specificity of PET and hCT in the ranges 0.70 to 0.90 and 0.80 to 0.90, respectively, with a power of 0.90 and a two-tailed 0.05 alpha.12 A sample size of 170 was estimated according to an expected prevalence of positive Ms-t in the range 0.20 to 0.24, but given the observed frequency of 0.28, just 132 patients were sufficient to collect the 36 positive Ms-t cases. The diagnostic estimates were sensitivity, specificity, and positive and negative predicted probabilities (PPP, NPP). To generalize the diagnostic estimates, the likelihood ratio for positive and negative results (LR+, LR-) are also calculated.
Database editing and statistical analyses were performed using SAS (SAS software 8.02 SAS Institute Inc, Cary, NC).
RESULTS
During the study period 675 lung cancer patients were attended at the HU12X. Of them 549 were NSCLC and 212 considered potentially operable. Of these, six rejected any medical procedure and 12 had extended disease and were kept out of surgery. Among the 194 eligible patients, eight did not consent to the study, 46 did not enter for logistical reasons, five because of hyperglycemia and three because of contrast allergies. The remaining 132 patients were referred to Ms-t. Thirty-six patients were Ms-t positive for mediastinal metastases, one had extensive invasion of local mediastinal structures, and 95 were Ms-t negative. Of these, three patients dropped out of the study because of a quick clinical deterioration, and 92 patients underwent thoracotomy. Thus, the index tests were compared to the Single Reference Standard in 132 patients and to the Composite Reference Standard in 129.
Table 1 shows some characteristics of the study subjects and compares them with the eligible patients that were not included. According to the definitive staging (either by mediastinoscopy or by thoracotomy), 39% of patients had a rather advanced disease at the time of diagnosis. No new mediastinal disease appeared in the 6 month follow-up.
The inter-observer agreement for the four levels classification hCTN0 to hCTN3 (quadratic weights) was 0.66 (95% CI, 0.54 to 0.79) and 0.56 (95% CI, 0.35 to 0.78) for hCTN0,1 versus hCTN2,3. The corresponding inter-rater agreements for PET were 0.78 (95% CI, 0.64 to 0.83) and 0.68 (95% CI, 0.55 to 0.80). No adverse effects attributable to the tests were found.
hCT identified 1,367 mediastinal lymph nodes (median 10, IQR 6-14) and 272 distinct nodal stations with enlarged lymph nodes; 164 of these stations were confined to the mediastinum of 63 patients. PET detected hot spots in 89 mediastinal stations of 52 patients. About two-thirds of patients with mediastinal involvement on either test had more than one station labeled as positive.
Ms-t systematically explored 717 mediastinal stations and found 78 invaded (median 2, IQR 1-3) in 36 patients, and all but one patient with one mediastinal station inaccessible to Ms-t (3, 8, or 9) had additional stations actually sampled by mediastinoscopy. No mortality or complications were associated with the procedure. Thoracotomy was undertaken in 92 patients, 253 lymph node stations (median 3, IQR 2-4) were sampled, and 26 of them in 22 patients had tumor invasion: station 5 in four patients, 7 and 9D in one patient, and 11 in 20 patients. Thoracotomy identified 10 additional cases of direct mediastinal invasion from the lung tumor (T4). Radical lymphadenectomy was not performed.
At the study closing, after a median follow-up of 42 months (IQR 36-49, range 30 to 55), 70 patients (51%) had died, and two patients had developed nodal mediastinal involvement at months 10 and 37, respectively.
Table 2 displays the diagnostic estimates of hCT and PET and shows that both tests perform similarly; actually, the hypothesis of no differences in their potential for classification cannot be rejected (McNemar’s test 2 0.973; P = .32). Given their high sensitivities, the main diagnostic value of both tests resides in the negative results that, for prevalences around 0.28, translates into high negative predicted probabilities. As a counterpart, the specificities are considerably low.
Regarding PET, five of the seven false-negative cases showed hot spots solely in the hilar-central station (10), while mediastinoscopy identified metastatic lymph nodes in stations 4, 5, 6 or 7, which neighbor station 10. Given its low anatomic resolution these results can be thought of as mediastinal (N2) cases misclassified by PET as hilar-central (N1c). To handle this potential for misclassification, a new threshold was set that classified PET results as positive (PETN1c,2,3) if the hot spot was allocated to the mediastinum or to the hilar- central station (10), and as negative (PETN0,1p) if it was assigned to the hilar-peripheral station (11). Compared with the former, this new threshold yields a sensitivity 10 points higher (0.94), a much lower specificity (0.59), and a negative likelihood ratio half in value (0.09).
In this particular case series, a positive test result for hCTN2,3, PETN2,3, or PETN1c,2,3 translates into positive predicted probabilities (PPP, true-positive) of 0.49, 0.56, and 0.47, respectively, that means the likelihood of submitting the patient to a futile thoracotomy; their complementary probabilities (1-PPP, false-positive) are 0.51, 0.44, and 0.53 and means the likelihood of erroneously omitting a potentially curative surgery. On the other hand, negative test results translate into negative predicted probabilities (NPP, true-negative) of 0.93, 0.91, and 0.97, respectively, that means the likelihood of properly submitting the patient for potential curative surgery (true-negative); their complementary probabilities (1-NPP, false-negative), are 0.07, 0.09, and 0.03, and means the likelihood of erroneously indicating resective thoracotomies.
The type of PET camera did not influence PET results. A logistic model was fitted with the variable PET results (negative v positive) regressing on Ms-t results (negative v positive). When the variable type of camera (POSICAN v ECAT EXACT 47) was added to the model its coefficient was not different from zero (P = .527), and the parameters of the univariable model did not change (likelihood ratio test 35.8236 and 36.2238) nor did the coefficient of the variable PET results (2.576 and 2.574).
PET and hCT results are conditionally dependent on each other. Table 3 shows the data of a stratified analysis of PET on hCT and vice versa, and the diagnostic estimates of each stratum. Although the numbers are small and the confidence intervals large, the table shows a trend for disclosing more PET false-positive results within large nodes (hCT positive) and more false-negatives within small ones. Accordingly, it can be said that PET achieves a more frequent correct identification of large benign nodes and of small malignant ones. The same applies to hCT with regard to PET.
Table 4 displays the diagnostic estimates of both tests performed in parallel (hCT and PET). At the high-threshold PET test (PETN2,3) sensitivity is 0.92 and 0.90 for the Single and Composite Reference Standards, respectively, and specificity 0.55 and 0.59; at the low-threshold (PETN1c,2,3) the sensitivity is 0.97 and 0.98, and the specificity 0.44 and 0.48 for both reference standards, respectively. It is important to note that the diagnostic estimates are quite similar for the two gold standards, and that, in this case series, they translate into high NPP, particularly with the low-threshold PET test (0.98; 95% CI, 0.88 to 1.00). Table 5 shows that the PPP varies widely across the different combinations of results of the component tests, depending on the prevalence of positive mediastinoscopy and on the operational characteristics of the tests within each subgroup. The combination of normal-sized lymph nodes in hCT and negative low-threshold PET test (PETN1c,2,3) gives up a NPP of 0.98 and a comfortable tight range (95% CI, 0.88 to 1.00) for the values that this population parameter is likely to take on the basis of the current sample data. The same cannot be said for the post-test probabilities of the remaining subgroups.
It has been pointed out that PET detects unexpected distant metastases in at least 10% of the potentially operable patients.13 The findings of our study, while anecdotal because the study was not designed with this aim, suggest that they would have had little impact in practice. Fifteen patients (11%) had extra-mediastinal hot spots, but seven patients were actually staged by mediastinoscopy as positive. The remaining eight patients were scheduled for thoracotomy. One patient developed overt chest wall infiltration before surgery, three patients had a blank thoracotomy, one patient died right after surgery, and the remaining three patients were alive after the 2 year follow-up.
DISCUSSION
There are several main strengths in this study. First, the study sample is a good representation of the eligible patient population. Second, the readings of the index and reference tests are strictly masked and independent. Third, the readings of hCT and PET are independent and reciprocally blinded, which allows for analyzing the diagnostic performance of hCT and PET both alone and combined. Finally, the reference standard has been complemented by a long and clinically significant follow-up period.
We are aware of two potential limitations of the study. First, the inter-observer agreement in identifying mediastinal lymph nodes of at least 10 mm the shortest diameter and hot spots within the mediastinum is lower than others reported. This nondifferential misclassification error is difficult to avoid with manual measurements and when placing a PET hot spot exactly within the mediastinum. The inter-rater agreement by consensus limits the underestimation effect on the diagnostic estimates but might affect their generalization. Second, the surgical gold standard might be considered incomplete given that thoracotomies were not performed with radical lymphadenectomy. However, on the basis of (1) the observed multilevel involvement of the mediastinum, on (2) the highly standardized performance of Ms-t, which systematically sampled every single accessible station, and on (3) the lack of clinical evidence of additional cases of mediastinal metastasis 6 months after thoracotomy, we believe that the effect of this limitation on the results must be small, if any. The prevalence of mediastinal nodal involvement both through mediastinoscopy and thoracotomy is similar (31% v 31.2%) to that found in the outstanding work by Pieterman14 in a series of 112 patients, where surgical resection was performed with radical lymphadenectomy. But the yield of the Ms-t is almost doubled in our study (28% v 15%). It is not surprising that the Composite Reference Standard adds little to the Single Reference Standard, which will be used as the standard of reference for the remaining part of this section. It can be said that whenever a thorough surgical exploration for mediastinal nodal metastasis is not done the risk of underestimating index tests specificity increases.
Table 6 summarizes the diagnostic estimates of six meta-analyses of a number of papers published between 1980 and 2002, and of four articles published afterwards.7,8,15-22 Most of the studies were conducted in the 1980s and 1990s, when the requirements for clinical studies on diagnostic performance were not as rigorous as they are now.23-25 They widely differ in terms of sample selection and size, in terms of the clinical and pathologic spectrum covered, and in the characteristics and undertaking of the index and reference tests, as well as in the way they were interpreted and compared. The potential for biased results26 in the original reports and the clinical and statistical heterogeneity27 in the meta-analyses are helpful to explain the different, and often conflicting, results observed.
In our study, the sensitivity (0.86) of hCT ranks among the highest reported, possibly because every patient underwent a contrast-enhanced spiral CT performed under a strict operating procedure, and after a conventional CT, as part of the clinical work-up. This strategy might have led to a selection effect, discarding patients with infiltration of the thoracic wall or vertebrae, and gross mediastinal invasion. Had this bias been absent, the sensitivity would have been higher. The specificity (0.67) ranks among the lowest described, giving a false-positive rate of 0.33. Twenty of the 32 false-positive patients were alive at the study closing, with no evidence of regional or distant metastasis after a median follow-up of 44 months (range, 30 to 55). Thus, the hCT false-positive rate would be somewhere between 0.33 (32 of 96 patients) and 0.21(20 of 96 patients).
At the high-threshold (PETN2,3) the sensitivity of PET is 0.81 (95% CI, 0.64 to 0.89), similar to that reported in many papers; However, at the low-threshold (PETN1c,2,3) the sensitivity of PET is 0.94 (95% CI, 0.81 to 0.91), which ranks among the highest described. Given that no radical lymphadenectomy was performed, the specificity (0,67 at the high-threshold and 0,59 at the low one) would be falsely low. Nonetheless, for the reasons given above, the magnitude of this bias is likely to be minimal. Furthermore, in the scenario of a low-threshold PET test 18 of the 39 false-positive patients were alive at the study closing, with no evidence of regional or distant metastasis after a median follow-up of 39 months (range, 30 to 51). Accordingly, the PET false-positive rate might be in the range of 0.41 (39 of 96 patients) and 0.19 (18 of 96 patients), which includes the value 0.23 (23 of 96 patients) for the high-threshold PET test.
As far as we know, there are three outcome studies that evaluate the clinical impact on unnecessary thoracotomies avoided by incorporating PET scans in staging NSCLC. Their conclusions are apparently contradictory although they differ in study design, outcome definition, and study performance.28-30
Our data show that any positive results on either test need to be cito-histologically confirmed before submitting the patient for surgery or chemoradiotherapy, and that negative results in hCT and PET decrease the post-test probabilities to levels that allow clinicians to reasonably rule out mediastinal involvement and to go straight to thoracotomy. Our personal recommendations, when performing a noninvasive NSCLC mediastinal staging, are to choose highly sensitive tests like the low-threshold FDG-PET (PETN1c,2,3) or the combined in-parallel hCT and PET. Given the outstanding anatomic resolution of CT and the dependency of clinicians and surgeons on it, hCT is likely to be used in all cases, and the question is whether a PET scan has to be performed on all patients regardless of the result of hCT. We believe that after a negative hCT, performing a PET scan is helpful because of the outstanding NPP yielded by a negative combined test. In the face of an abnormal hCT a negative PET result seems to provide a potential diagnostic benefit by increasing to 0,94 the probability of properly missing a thoracotomy. Unfortunately, this estimate is based in small numbers, lacks statistical precision (95% CI, 0.70 to 0.99), and consequently, does not allow us to confidently advise physicians to start any treatment without histopathologic confirmation. The integrated PET and CT scans,31 while improving the anatomic location of hot spots, cannot at present be a substitute for the exquisite anatomic information derived from the high definition of hCT.
Further studies are needed to place these statements into an economic perspective and to find out the combination of tests with the greatest impact on clinical outcomes.
Authors’ Disclosures of Potential Conflicts of Interest
Although all authors completed the disclosure declaration, the following author or immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed discription of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
NOTES
Supported by a research grant (FIS 98/0702) from the Fund for Health Research and Red Respira (RTIC 03/11), Ministry of Health, Spain. The Centro PET Complutense was paid for all PET scans by the Hospital Universitario 12 de Octubre.
Presented in part at the IX Annual Meeting of the Society of Pneumology and Thoracic Surgery of Madrid, Madrid, Spain, April 2004; XXXVII Annual Meeting of the Spanish Society of Pneumology and Thoracic Surgery, Madrid, Spain, June 2004; 25th Annual Meeting of the Spanish Society of Nuclear Medicine, Barcelona, Spain, June 2004; 51st Annual Meeting of the Society of Nuclear Medicine, Philadelphia, PA, June 2004; XIV Annual Conference of the European Respiratory Society, Glasgow, UK, September 2004.
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
European Network of Cancer Registry. Lung cancer in Europe. Jerzy E. Tyczynsky, Freddie Bray, D Maxwell Parkin. 5/2004 update. http://www.encr.com
Naruke T, Tsuchiya R, Kondo H, et al: Prognosis and survival after resection for bronchogenic carcinoma based on the 1997 TNM-staging classification: The Japanese experience. Ann Thorac Surg 71:1759-1764, 2001
Mortalidad por cáncer en Espa?a. área de Epidemiología Ambiental y Cáncer. Centro Nacional de Epidemiología. Instituto de Salud Carlos III, Ministry of Health, Spain. 5/2004 update. http://cne.isciii.es/cancer/mort2001.txt .
Downey RJ: Follow-up of patients with completely resected lung cancer. Chest 115:1487-1498, 1999
Dillemans B, Deneffe G, Verschakelen J, et al: Value of computed tomography and mediastinoscopy in preoperative evaluation of mediastinal nodes in non-small cell lung cancer. A study of 569 patients. Eur J Cardiothorac Surg 8:37-42, 1994
Toloza EM, Harpole L, Detterbeck F, et al: Invasive staging of non-small cell lung cancer: A review of the current evidence. Chest 123:S157–S166, 2003
Dales RE, Stark RM, Raman S: Computed tomography to stage lung cancer. Approaching a controversy using meta-analysis. Am Rev Respir Dis 141:1096-1101, 1990
Dwamena BA, Sonnad SS, Angobaldo JO, et al: Metastases from non-small cell lung cancer: Mediastinal staging in the 1990s–meta-analytic comparison of PET and CT. Radiology 213:530-536, 1999
Pautas, Carcinoma Broncogénico. Hospital Universitario 12 de Octubre, Madrid, Spain López-Encuentra A, and Grupo Cooperativo del Carcinoma Broncogénico. 5/2004 update. http://www.mbeneumologia.org/mbe/pautas.aspx
Mountain CF, Dresler CM: Regional lymph node classification for lung cancer staging. Chest 111:1718-1723, 1997
Hamacher K, Coenen HH, Stocklin G: Efficient stereospecific synthesis of no-carrier-added 2-[18F]-fluoro-2-deoxy-D-glucose using aminopolyether supported nucleophilic substitution. J Nucl Med 27:235-238, 1986
Arkin CF, Wachtel MS: How many patients are necessary to assess test performance? JAMA 263:275-278, 1990
Weder W, Schmid RA, Bruchhaus H, et al: Detection of extrathoracic metastases by positron emission tomography in lung cancer. Ann Thorac Surg 66:886-892, 1998
Pieterman RM, van Putten JW, Meuzelaar JJ, et al: Preoperative staging of non-small-cell lung cancer with positron-emission tomography. N Engl J Med 343:254-261, 2000
Hellwig D, Ukena D, Paulsen F, et al: [Meta-analysis of the efficacy of positron emission tomography with F-18-fluorodeoxyglucose in lung tumors. Basis for discussion of the German Consensus Conference on PET in Oncology 2000]. Pneumologie 55:367-377, 2001
Toloza EM, Harpole L, McCrory DC: Noninvasive staging of non-small cell lung cancer: A systematic review of the current evidence. Chest 123:S137–S146, 2003
Positron emission tomography (PET) imaging in cancer management. Health Technology Assessment Report 2. Glasgow: Health Technology Board for Scotland. Bradbury I, Bonell E, Boynton J, et al: 2002. Update 6/2003. www.htbs.co.uk
Gould MK, Kuschner WG, Rydzak CE, et al: Test performance of positron emission tomography and computed tomography for mediastinal staging in patients with non-small-cell lung cancer: A meta-analysis. Ann Intern Med 139:879-892, 2003
Gonzalez-Stawinski GV, Lemaire A, Merchant F, et al: A comparative analysis of positron emission tomography and mediastinoscopy in staging non-small cell lung cancer. J Thorac Cardiovasc Surg 126:1900-1905, 2003
Cerfolio RJ, Ojha B, Bryant AS, et al: The role of FDG-PET scan in staging patients with nonsmall cell carcinoma. Ann Thorac Surg 76:861-866, 2003
Verhagen AF, Bootsma GP, Tjan-Heijnen VC, et al: FDG-PET in staging lung cancer; How does it change the algorithm? Lung Cancer 44:175-181, 2004
Kelly RF, Tran T, Holmstrom A, et al: Accuracy and cost-effectiveness of [18F]-2-fluoro-deoxy-D-glucose-positron emission tomography scan in potentially resectable non-small cell lung cancer. Chest 125:1413-1423, 2004
Sheps SB, Schechter MT: The assessment of diagnostic tests. A survey of current medical research. JAMA 252:2418-2422, 1984
Reid MC, Lachs MS, Feinstein AR: Use of methodological standards in diagnostic test research. Getting better but still not good. JAMA 274:645-651, 1995
Bossuyt PM, Reitsma JB, Bruns DE, et al: Towards complete and accurate reporting of studies of diagnostic accuracy: The STARD initiative. BMJ 326:41-44, 2003
Begg CB: Biases in the assessment of diagnostic tests. Stat Med 6:411-423, 1987
Lijmer JG, Bossuyt PM, Heisterkamp SH: Exploring sources of heterogeneity in systematic reviews of diagnostic tests. Stat Med 21:1525-1537, 2002
Van Tinteren H, Hoekstra OS, Smit EF, et al: Effectiveness of positron emission tomography in the preoperative assessment of patients with suspected non-small-cell lung cancer: The PLUS multicentre randomised trial. Lancet 359:1388-1393, 2002
Reed CE, Harpole DH, Posther KE, et al: Results of the American College of Surgeons Oncology Group Z0050 trial: The utility of positron emission tomography in staging potentially operable non-small cell lung cancer. J Thorac Cardiovasc Surg 126:1943-1951, 2003
Viney RC, Boyer MJ, King MT, et al: Randomized controlled trial of the role of positron emission tomography in the management of stage I and II non-small-cell lung cancer. J Clin Oncol 22:2357-2362, 2004
Schirrmeister H, Hetzel M, Buck A: Staging of non-small-cell lung cancer with integrated PET and CT. N Engl J Med 349:1188-1190, 2003(Francisco Pozo-Rodríguez,)
ABSTRACT
PURPOSE: Computed tomography (CT) and [18F] Fluorodeoxyglucose positron emission tomography (FDG-PET) are considered suitable methods for the noninvasive staging of the mediastinum. Our study was intended to estimate the efficacy of contrast-enhanced helical CT (hCT) and FDG-PET, alone and combined, in the diagnosis of lymph node mediastinal metastases.
METHODS: This study was a prospective and blind comparison of the efficacy of hCT and FDG-PET with two alternative reference standards, mediastinoscopy, and mediastinoscopy plus thoracotomy plus a 6-month follow-up to diagnose lymph node mediastinal metastases in 132 consecutive patients with potentially resectable non–small-cell lung cancer (NSCLC). The metastatic disease was assessed histopathologically. Further clinical information was obtained postoperatively after a median follow-up of 42 months.
RESULTS: The prevalence of cN2,3 is 0.28. For hCT the sensitivity and specificity are 0.86 (95% CI, 0.70 to 0.93) and 0.67 (95% CI, 0.56 to 0.75), for PET 0.94 (95% CI, 0.81 to 0.98) and 0.59 (95% CI, 0.49 to 0.68), and for hCT and PET combined in-parallel 0.97 (95% CI, 0.84 to 0.99) and 0.44 (95% CI, 0.34 to 0.53), which translate into a negative predicted probability of 0.98 (95% CI, 0.88 to 1.00). The crude diagnostic odds ratio of PET in the total sample studied is 13.1, in the subgroup hCT+ 11.04 (3.0 to 40 0.1), and in the hCT- 3.5 (0.5 to 21.5). Similar results were obtained for hCT stratified by PET.
CONCLUSION: hCT and PET perform similarly in the mediastinal staging of NSCLC, both tests are conditionally dependent and provide complementary information, and their diagnostic value mainly resides on the negative results.
INTRODUCTION
Lung cancer is the most common form of cancer.1 In the year 2000 the crude incidence rate in Spain was 86.7/105 in men and 7.7/105 in women. Lung cancer is a lethal disease with a 5-year survival rate of 15% overall and around 40% for those operated, that varies with the extent of the disease.2 In the year 2000 the Spanish National Center of Epidemiology reported 15,432 and 1,876 male and female deaths from lung cancer, an age adjusted mortality rate of 48.23/105 and 4.85/105, and a lifetime risk of dying of 5.56% and 0.51%, respectively.3
Surgical resection is the accepted standard of care for patients with non–small-cell lung cancer (NSCLC), but only approximately 25% are eligible for surgery.4 Once the surgical candidates are selected mediastinal staging is mandatory because up to 50% still have regional metastasis.5 Staging involves a considerable effort in terms of number of tests, duration, and cost.
The standard for invasive staging is mediastinoscopy/tomy (Ms-t).6 Computed tomography (CT) came up in the late 1970s as a noninvasive suitable staging diagnostic technique, but the papers published since then have shown conflicting results, and a meta-analysis published in 1990 concluded that no additional significant advances were to be expected from CT.7 In the early 1990s, positron emission tomography with [18F] Fluorodeoxyglucose (FDG-PET) emerged as a promising staging method because of its ability to detect areas of high FDG metabolic activity consistent with malignancy. The published reports generally conclude that PET is superior to CT for the mediastinum and point out the potential of PET to detect unexpected distant metastases.8 They rarely inform of how CT and PET perform together.
The goal of the present study is to determine the diagnostic efficacy of the helical-enhanced CT of the chest and upper abdomen (hCT) and the whole–body FDG–PET, used alone and combined in parallel, in the mediastinal staging of potentially operable NSCLC patients.
METHODS
Study Design
This is an observational study, cross-sectional with concurrently gathered information, performed between February 1999 and May 2001 at the Hospital Universitario 12 de Octubre (HU12X), a tertiary level hospital that provides health care for a population of 750,000 in Madrid. The target condition is metastatic disease in mediastinal lymph nodes, the index tests are hCT and PET, alone and combined in parallel; there are two reference standards: mediastinoscopy (Single Reference Standard-SRS) and mediastinoscopy plus thoracotomy plus a 6-month follow-up (Composite Reference Standard-CRS).
The study was designed to mimic the standard of clinical practice at the HU12X, having hCT and PET scans performed concurrently and apart from the standard clinical work-up and keeping the clinical decision makers unaware of their results. Both tests had to be done within 72 hours before the scheduled mediastinoscopy, and Ms-t had to be performed by the same two surgeons; otherwise, the patient did not enter the study for reasons that will be denominated as logistical. Radiologists and nuclear medicine physicians were blinded to clinical information and to PET and hCT scans, respectively. Two independent readings were obtained from each test and discrepancies solved by consensus. Following the study design, all clinical decisions were made after the patient was given the current standard of clinical practice and staged accordingly; PET or hCT images that suggested systemic metastases neither were notified to clinicians nor had to be histologically confirmed. The study was approved by the Ethics Committee of the HU12X, and the study subjects were presented with a form for written informed consent.
Participants
The eligible subjects were patients aged over 17, with histologically diagnosed NSCLC, staged I, II, and selected III, and potentially resectable. The exclusion criteria were allergy to iodine contrast and hyperglycemia over 9 mmol/L on the day PET was performed. The study subjects were all the consecutive eligible patients referred for Ms-t during the study period, who did not meet any of the exclusion criteria and gave their written informed consent.
Study Procedures
According to the standard of clinical practice of the HU12X, all patients histologically diagnosed from NSCLC underwent a study of operability consisting of a complete medical history and physical examination, routine hematologic and biochemical analyses, chest x-ray, ECG, bronchoscopy, conventional CT of the chest and upper abdomen, and lung function tests. Pertinent tests were performed whenever distant metastases were suspected. All potentially operable patients were subjected to cervical mediastinoscopy with parasternal mediastinotomy, if the tumor settled on the left upper lobe or left main bronchus. The Ms-t negative cases were referred to thoracotomy for pulmonary resection and mediastinal sampling.9 Staging was done according to the revised 1997 International System for Staging Lung Cancer.10
In addition to and apart from the standard work-up, the study patients undertook both an hCT and a whole-body PET scan. The hCT was performed with a 120 to 140 KV, 200 to 250 mA Toshiba X-Press spiral scan (Toshiba Medical Systems, Tokyo, Japan). Sections 7 mm thick from supraclavicular to infrarenal regions were obtained with a pitch 1,4 and a reconstruction interval of 5 mm. Each patient was administered intravenously 100 cm3 of a 0.117 mmol/cm3 iodine contrast solution (Opitray 300, Tyco Healthcare) at a rate of 2 cm3/s. There was a time lag of 30 seconds between the start of the perfusion and commencement of the scanning. The information gathered was the number and size of mediastinal lymph nodes detected. An hCT was considered positive (hCTN2,3) if the nodal shortest diameter was 10 mm or larger, and negative otherwise (hCTN0,1).
PET was performed on all patients at the Centro PET Complutense (Madrid), a private facility and the first PET center installed in Spain (1995). The first 47 scans were done with a POSICAM HZL-R camera (Positron Corporation, Houston, TX) with a field of view of 15 cm, a full width at half-maximal resolution of 5 mm, and correction for attenuation in thorax. The remaining 85 patients were scanned with an ECAT EXACT 47 camera (Siemens-CTI, Erlangen, Germany) with whole body attenuation, a field of view of 16.2 cm, a spatial resolution of 4.5 (2D), and 4.6 mm (3D). 18FDG was synthesized at the PET Center in a 12 Mev cyclotron (Oxford Instruments, Oxford, United Kingdom), following Hamacher's method.11 Examinations were performed after 6 hours of fasting. The transmission study (slices of width 16.6 cm) was done with a germanium-68 source during 20 minutes. Then, the patient was injected an intravenous dose of 370 MBq 18FDG. The emission study commenced 45 minutes after the injection and lasted for 20 minutes. Slices of width 16.2 cm were subsequently obtained with a 5 minutes exposition each, covering from the first cervical vertebra to the groins and corrected for attenuation with a Sun Ultra 60 computer 2 x 450 Mhz with an iterative reconstruction type weighted attenuation OSEM. Results were considered positive (PETN,2,3) when at least one definite, localized area of higher uptake than the mediastinal background was present, and negative otherwise (PETN0,1).
The result of the hCT and PET combined (in parallel) test was considered positive if any one of the components was positive and negative if they were both negative.
All mediastinoscopies were performed by the same two surgeons, who did a systematic sampling of normally and abnormally appearing mediastinal lymph nodes and fat. Stations 2, 4-superior and 4-inferior (left and right), and 7 were explored via cervical mediastinoscopy, and stations 5 and 6 via mediastinotomy. Thoracotomies were performed by seven thoracic surgeons. The samples obtained were stained with hematoxylin-eosin and examined histopathologically. Results were positive whenever metastases to lymph nodes were present.
Statistical Analysis
A minimum of 36 positive Ms-ts were needed to detect a sensitivity and specificity of PET and hCT in the ranges 0.70 to 0.90 and 0.80 to 0.90, respectively, with a power of 0.90 and a two-tailed 0.05 alpha.12 A sample size of 170 was estimated according to an expected prevalence of positive Ms-t in the range 0.20 to 0.24, but given the observed frequency of 0.28, just 132 patients were sufficient to collect the 36 positive Ms-t cases. The diagnostic estimates were sensitivity, specificity, and positive and negative predicted probabilities (PPP, NPP). To generalize the diagnostic estimates, the likelihood ratio for positive and negative results (LR+, LR-) are also calculated.
Database editing and statistical analyses were performed using SAS (SAS software 8.02 SAS Institute Inc, Cary, NC).
RESULTS
During the study period 675 lung cancer patients were attended at the HU12X. Of them 549 were NSCLC and 212 considered potentially operable. Of these, six rejected any medical procedure and 12 had extended disease and were kept out of surgery. Among the 194 eligible patients, eight did not consent to the study, 46 did not enter for logistical reasons, five because of hyperglycemia and three because of contrast allergies. The remaining 132 patients were referred to Ms-t. Thirty-six patients were Ms-t positive for mediastinal metastases, one had extensive invasion of local mediastinal structures, and 95 were Ms-t negative. Of these, three patients dropped out of the study because of a quick clinical deterioration, and 92 patients underwent thoracotomy. Thus, the index tests were compared to the Single Reference Standard in 132 patients and to the Composite Reference Standard in 129.
Table 1 shows some characteristics of the study subjects and compares them with the eligible patients that were not included. According to the definitive staging (either by mediastinoscopy or by thoracotomy), 39% of patients had a rather advanced disease at the time of diagnosis. No new mediastinal disease appeared in the 6 month follow-up.
The inter-observer agreement for the four levels classification hCTN0 to hCTN3 (quadratic weights) was 0.66 (95% CI, 0.54 to 0.79) and 0.56 (95% CI, 0.35 to 0.78) for hCTN0,1 versus hCTN2,3. The corresponding inter-rater agreements for PET were 0.78 (95% CI, 0.64 to 0.83) and 0.68 (95% CI, 0.55 to 0.80). No adverse effects attributable to the tests were found.
hCT identified 1,367 mediastinal lymph nodes (median 10, IQR 6-14) and 272 distinct nodal stations with enlarged lymph nodes; 164 of these stations were confined to the mediastinum of 63 patients. PET detected hot spots in 89 mediastinal stations of 52 patients. About two-thirds of patients with mediastinal involvement on either test had more than one station labeled as positive.
Ms-t systematically explored 717 mediastinal stations and found 78 invaded (median 2, IQR 1-3) in 36 patients, and all but one patient with one mediastinal station inaccessible to Ms-t (3, 8, or 9) had additional stations actually sampled by mediastinoscopy. No mortality or complications were associated with the procedure. Thoracotomy was undertaken in 92 patients, 253 lymph node stations (median 3, IQR 2-4) were sampled, and 26 of them in 22 patients had tumor invasion: station 5 in four patients, 7 and 9D in one patient, and 11 in 20 patients. Thoracotomy identified 10 additional cases of direct mediastinal invasion from the lung tumor (T4). Radical lymphadenectomy was not performed.
At the study closing, after a median follow-up of 42 months (IQR 36-49, range 30 to 55), 70 patients (51%) had died, and two patients had developed nodal mediastinal involvement at months 10 and 37, respectively.
Table 2 displays the diagnostic estimates of hCT and PET and shows that both tests perform similarly; actually, the hypothesis of no differences in their potential for classification cannot be rejected (McNemar’s test 2 0.973; P = .32). Given their high sensitivities, the main diagnostic value of both tests resides in the negative results that, for prevalences around 0.28, translates into high negative predicted probabilities. As a counterpart, the specificities are considerably low.
Regarding PET, five of the seven false-negative cases showed hot spots solely in the hilar-central station (10), while mediastinoscopy identified metastatic lymph nodes in stations 4, 5, 6 or 7, which neighbor station 10. Given its low anatomic resolution these results can be thought of as mediastinal (N2) cases misclassified by PET as hilar-central (N1c). To handle this potential for misclassification, a new threshold was set that classified PET results as positive (PETN1c,2,3) if the hot spot was allocated to the mediastinum or to the hilar- central station (10), and as negative (PETN0,1p) if it was assigned to the hilar-peripheral station (11). Compared with the former, this new threshold yields a sensitivity 10 points higher (0.94), a much lower specificity (0.59), and a negative likelihood ratio half in value (0.09).
In this particular case series, a positive test result for hCTN2,3, PETN2,3, or PETN1c,2,3 translates into positive predicted probabilities (PPP, true-positive) of 0.49, 0.56, and 0.47, respectively, that means the likelihood of submitting the patient to a futile thoracotomy; their complementary probabilities (1-PPP, false-positive) are 0.51, 0.44, and 0.53 and means the likelihood of erroneously omitting a potentially curative surgery. On the other hand, negative test results translate into negative predicted probabilities (NPP, true-negative) of 0.93, 0.91, and 0.97, respectively, that means the likelihood of properly submitting the patient for potential curative surgery (true-negative); their complementary probabilities (1-NPP, false-negative), are 0.07, 0.09, and 0.03, and means the likelihood of erroneously indicating resective thoracotomies.
The type of PET camera did not influence PET results. A logistic model was fitted with the variable PET results (negative v positive) regressing on Ms-t results (negative v positive). When the variable type of camera (POSICAN v ECAT EXACT 47) was added to the model its coefficient was not different from zero (P = .527), and the parameters of the univariable model did not change (likelihood ratio test 35.8236 and 36.2238) nor did the coefficient of the variable PET results (2.576 and 2.574).
PET and hCT results are conditionally dependent on each other. Table 3 shows the data of a stratified analysis of PET on hCT and vice versa, and the diagnostic estimates of each stratum. Although the numbers are small and the confidence intervals large, the table shows a trend for disclosing more PET false-positive results within large nodes (hCT positive) and more false-negatives within small ones. Accordingly, it can be said that PET achieves a more frequent correct identification of large benign nodes and of small malignant ones. The same applies to hCT with regard to PET.
Table 4 displays the diagnostic estimates of both tests performed in parallel (hCT and PET). At the high-threshold PET test (PETN2,3) sensitivity is 0.92 and 0.90 for the Single and Composite Reference Standards, respectively, and specificity 0.55 and 0.59; at the low-threshold (PETN1c,2,3) the sensitivity is 0.97 and 0.98, and the specificity 0.44 and 0.48 for both reference standards, respectively. It is important to note that the diagnostic estimates are quite similar for the two gold standards, and that, in this case series, they translate into high NPP, particularly with the low-threshold PET test (0.98; 95% CI, 0.88 to 1.00). Table 5 shows that the PPP varies widely across the different combinations of results of the component tests, depending on the prevalence of positive mediastinoscopy and on the operational characteristics of the tests within each subgroup. The combination of normal-sized lymph nodes in hCT and negative low-threshold PET test (PETN1c,2,3) gives up a NPP of 0.98 and a comfortable tight range (95% CI, 0.88 to 1.00) for the values that this population parameter is likely to take on the basis of the current sample data. The same cannot be said for the post-test probabilities of the remaining subgroups.
It has been pointed out that PET detects unexpected distant metastases in at least 10% of the potentially operable patients.13 The findings of our study, while anecdotal because the study was not designed with this aim, suggest that they would have had little impact in practice. Fifteen patients (11%) had extra-mediastinal hot spots, but seven patients were actually staged by mediastinoscopy as positive. The remaining eight patients were scheduled for thoracotomy. One patient developed overt chest wall infiltration before surgery, three patients had a blank thoracotomy, one patient died right after surgery, and the remaining three patients were alive after the 2 year follow-up.
DISCUSSION
There are several main strengths in this study. First, the study sample is a good representation of the eligible patient population. Second, the readings of the index and reference tests are strictly masked and independent. Third, the readings of hCT and PET are independent and reciprocally blinded, which allows for analyzing the diagnostic performance of hCT and PET both alone and combined. Finally, the reference standard has been complemented by a long and clinically significant follow-up period.
We are aware of two potential limitations of the study. First, the inter-observer agreement in identifying mediastinal lymph nodes of at least 10 mm the shortest diameter and hot spots within the mediastinum is lower than others reported. This nondifferential misclassification error is difficult to avoid with manual measurements and when placing a PET hot spot exactly within the mediastinum. The inter-rater agreement by consensus limits the underestimation effect on the diagnostic estimates but might affect their generalization. Second, the surgical gold standard might be considered incomplete given that thoracotomies were not performed with radical lymphadenectomy. However, on the basis of (1) the observed multilevel involvement of the mediastinum, on (2) the highly standardized performance of Ms-t, which systematically sampled every single accessible station, and on (3) the lack of clinical evidence of additional cases of mediastinal metastasis 6 months after thoracotomy, we believe that the effect of this limitation on the results must be small, if any. The prevalence of mediastinal nodal involvement both through mediastinoscopy and thoracotomy is similar (31% v 31.2%) to that found in the outstanding work by Pieterman14 in a series of 112 patients, where surgical resection was performed with radical lymphadenectomy. But the yield of the Ms-t is almost doubled in our study (28% v 15%). It is not surprising that the Composite Reference Standard adds little to the Single Reference Standard, which will be used as the standard of reference for the remaining part of this section. It can be said that whenever a thorough surgical exploration for mediastinal nodal metastasis is not done the risk of underestimating index tests specificity increases.
Table 6 summarizes the diagnostic estimates of six meta-analyses of a number of papers published between 1980 and 2002, and of four articles published afterwards.7,8,15-22 Most of the studies were conducted in the 1980s and 1990s, when the requirements for clinical studies on diagnostic performance were not as rigorous as they are now.23-25 They widely differ in terms of sample selection and size, in terms of the clinical and pathologic spectrum covered, and in the characteristics and undertaking of the index and reference tests, as well as in the way they were interpreted and compared. The potential for biased results26 in the original reports and the clinical and statistical heterogeneity27 in the meta-analyses are helpful to explain the different, and often conflicting, results observed.
In our study, the sensitivity (0.86) of hCT ranks among the highest reported, possibly because every patient underwent a contrast-enhanced spiral CT performed under a strict operating procedure, and after a conventional CT, as part of the clinical work-up. This strategy might have led to a selection effect, discarding patients with infiltration of the thoracic wall or vertebrae, and gross mediastinal invasion. Had this bias been absent, the sensitivity would have been higher. The specificity (0.67) ranks among the lowest described, giving a false-positive rate of 0.33. Twenty of the 32 false-positive patients were alive at the study closing, with no evidence of regional or distant metastasis after a median follow-up of 44 months (range, 30 to 55). Thus, the hCT false-positive rate would be somewhere between 0.33 (32 of 96 patients) and 0.21(20 of 96 patients).
At the high-threshold (PETN2,3) the sensitivity of PET is 0.81 (95% CI, 0.64 to 0.89), similar to that reported in many papers; However, at the low-threshold (PETN1c,2,3) the sensitivity of PET is 0.94 (95% CI, 0.81 to 0.91), which ranks among the highest described. Given that no radical lymphadenectomy was performed, the specificity (0,67 at the high-threshold and 0,59 at the low one) would be falsely low. Nonetheless, for the reasons given above, the magnitude of this bias is likely to be minimal. Furthermore, in the scenario of a low-threshold PET test 18 of the 39 false-positive patients were alive at the study closing, with no evidence of regional or distant metastasis after a median follow-up of 39 months (range, 30 to 51). Accordingly, the PET false-positive rate might be in the range of 0.41 (39 of 96 patients) and 0.19 (18 of 96 patients), which includes the value 0.23 (23 of 96 patients) for the high-threshold PET test.
As far as we know, there are three outcome studies that evaluate the clinical impact on unnecessary thoracotomies avoided by incorporating PET scans in staging NSCLC. Their conclusions are apparently contradictory although they differ in study design, outcome definition, and study performance.28-30
Our data show that any positive results on either test need to be cito-histologically confirmed before submitting the patient for surgery or chemoradiotherapy, and that negative results in hCT and PET decrease the post-test probabilities to levels that allow clinicians to reasonably rule out mediastinal involvement and to go straight to thoracotomy. Our personal recommendations, when performing a noninvasive NSCLC mediastinal staging, are to choose highly sensitive tests like the low-threshold FDG-PET (PETN1c,2,3) or the combined in-parallel hCT and PET. Given the outstanding anatomic resolution of CT and the dependency of clinicians and surgeons on it, hCT is likely to be used in all cases, and the question is whether a PET scan has to be performed on all patients regardless of the result of hCT. We believe that after a negative hCT, performing a PET scan is helpful because of the outstanding NPP yielded by a negative combined test. In the face of an abnormal hCT a negative PET result seems to provide a potential diagnostic benefit by increasing to 0,94 the probability of properly missing a thoracotomy. Unfortunately, this estimate is based in small numbers, lacks statistical precision (95% CI, 0.70 to 0.99), and consequently, does not allow us to confidently advise physicians to start any treatment without histopathologic confirmation. The integrated PET and CT scans,31 while improving the anatomic location of hot spots, cannot at present be a substitute for the exquisite anatomic information derived from the high definition of hCT.
Further studies are needed to place these statements into an economic perspective and to find out the combination of tests with the greatest impact on clinical outcomes.
Authors’ Disclosures of Potential Conflicts of Interest
Although all authors completed the disclosure declaration, the following author or immediate family members indicated a financial interest. No conflict exists for drugs or devices used in a study if they are not being evaluated as part of the investigation. For a detailed discription of the disclosure categories, or for more information about ASCO's conflict of interest policy, please refer to the Author Disclosure Declaration and the Disclosures of Potential Conflicts of Interest section in Information for Contributors.
NOTES
Supported by a research grant (FIS 98/0702) from the Fund for Health Research and Red Respira (RTIC 03/11), Ministry of Health, Spain. The Centro PET Complutense was paid for all PET scans by the Hospital Universitario 12 de Octubre.
Presented in part at the IX Annual Meeting of the Society of Pneumology and Thoracic Surgery of Madrid, Madrid, Spain, April 2004; XXXVII Annual Meeting of the Spanish Society of Pneumology and Thoracic Surgery, Madrid, Spain, June 2004; 25th Annual Meeting of the Spanish Society of Nuclear Medicine, Barcelona, Spain, June 2004; 51st Annual Meeting of the Society of Nuclear Medicine, Philadelphia, PA, June 2004; XIV Annual Conference of the European Respiratory Society, Glasgow, UK, September 2004.
Authors’ disclosures of potential conflicts of interest are found at the end of this article.
REFERENCES
European Network of Cancer Registry. Lung cancer in Europe. Jerzy E. Tyczynsky, Freddie Bray, D Maxwell Parkin. 5/2004 update. http://www.encr.com
Naruke T, Tsuchiya R, Kondo H, et al: Prognosis and survival after resection for bronchogenic carcinoma based on the 1997 TNM-staging classification: The Japanese experience. Ann Thorac Surg 71:1759-1764, 2001
Mortalidad por cáncer en Espa?a. área de Epidemiología Ambiental y Cáncer. Centro Nacional de Epidemiología. Instituto de Salud Carlos III, Ministry of Health, Spain. 5/2004 update. http://cne.isciii.es/cancer/mort2001.txt .
Downey RJ: Follow-up of patients with completely resected lung cancer. Chest 115:1487-1498, 1999
Dillemans B, Deneffe G, Verschakelen J, et al: Value of computed tomography and mediastinoscopy in preoperative evaluation of mediastinal nodes in non-small cell lung cancer. A study of 569 patients. Eur J Cardiothorac Surg 8:37-42, 1994
Toloza EM, Harpole L, Detterbeck F, et al: Invasive staging of non-small cell lung cancer: A review of the current evidence. Chest 123:S157–S166, 2003
Dales RE, Stark RM, Raman S: Computed tomography to stage lung cancer. Approaching a controversy using meta-analysis. Am Rev Respir Dis 141:1096-1101, 1990
Dwamena BA, Sonnad SS, Angobaldo JO, et al: Metastases from non-small cell lung cancer: Mediastinal staging in the 1990s–meta-analytic comparison of PET and CT. Radiology 213:530-536, 1999
Pautas, Carcinoma Broncogénico. Hospital Universitario 12 de Octubre, Madrid, Spain López-Encuentra A, and Grupo Cooperativo del Carcinoma Broncogénico. 5/2004 update. http://www.mbeneumologia.org/mbe/pautas.aspx
Mountain CF, Dresler CM: Regional lymph node classification for lung cancer staging. Chest 111:1718-1723, 1997
Hamacher K, Coenen HH, Stocklin G: Efficient stereospecific synthesis of no-carrier-added 2-[18F]-fluoro-2-deoxy-D-glucose using aminopolyether supported nucleophilic substitution. J Nucl Med 27:235-238, 1986
Arkin CF, Wachtel MS: How many patients are necessary to assess test performance? JAMA 263:275-278, 1990
Weder W, Schmid RA, Bruchhaus H, et al: Detection of extrathoracic metastases by positron emission tomography in lung cancer. Ann Thorac Surg 66:886-892, 1998
Pieterman RM, van Putten JW, Meuzelaar JJ, et al: Preoperative staging of non-small-cell lung cancer with positron-emission tomography. N Engl J Med 343:254-261, 2000
Hellwig D, Ukena D, Paulsen F, et al: [Meta-analysis of the efficacy of positron emission tomography with F-18-fluorodeoxyglucose in lung tumors. Basis for discussion of the German Consensus Conference on PET in Oncology 2000]. Pneumologie 55:367-377, 2001
Toloza EM, Harpole L, McCrory DC: Noninvasive staging of non-small cell lung cancer: A systematic review of the current evidence. Chest 123:S137–S146, 2003
Positron emission tomography (PET) imaging in cancer management. Health Technology Assessment Report 2. Glasgow: Health Technology Board for Scotland. Bradbury I, Bonell E, Boynton J, et al: 2002. Update 6/2003. www.htbs.co.uk
Gould MK, Kuschner WG, Rydzak CE, et al: Test performance of positron emission tomography and computed tomography for mediastinal staging in patients with non-small-cell lung cancer: A meta-analysis. Ann Intern Med 139:879-892, 2003
Gonzalez-Stawinski GV, Lemaire A, Merchant F, et al: A comparative analysis of positron emission tomography and mediastinoscopy in staging non-small cell lung cancer. J Thorac Cardiovasc Surg 126:1900-1905, 2003
Cerfolio RJ, Ojha B, Bryant AS, et al: The role of FDG-PET scan in staging patients with nonsmall cell carcinoma. Ann Thorac Surg 76:861-866, 2003
Verhagen AF, Bootsma GP, Tjan-Heijnen VC, et al: FDG-PET in staging lung cancer; How does it change the algorithm? Lung Cancer 44:175-181, 2004
Kelly RF, Tran T, Holmstrom A, et al: Accuracy and cost-effectiveness of [18F]-2-fluoro-deoxy-D-glucose-positron emission tomography scan in potentially resectable non-small cell lung cancer. Chest 125:1413-1423, 2004
Sheps SB, Schechter MT: The assessment of diagnostic tests. A survey of current medical research. JAMA 252:2418-2422, 1984
Reid MC, Lachs MS, Feinstein AR: Use of methodological standards in diagnostic test research. Getting better but still not good. JAMA 274:645-651, 1995
Bossuyt PM, Reitsma JB, Bruns DE, et al: Towards complete and accurate reporting of studies of diagnostic accuracy: The STARD initiative. BMJ 326:41-44, 2003
Begg CB: Biases in the assessment of diagnostic tests. Stat Med 6:411-423, 1987
Lijmer JG, Bossuyt PM, Heisterkamp SH: Exploring sources of heterogeneity in systematic reviews of diagnostic tests. Stat Med 21:1525-1537, 2002
Van Tinteren H, Hoekstra OS, Smit EF, et al: Effectiveness of positron emission tomography in the preoperative assessment of patients with suspected non-small-cell lung cancer: The PLUS multicentre randomised trial. Lancet 359:1388-1393, 2002
Reed CE, Harpole DH, Posther KE, et al: Results of the American College of Surgeons Oncology Group Z0050 trial: The utility of positron emission tomography in staging potentially operable non-small cell lung cancer. J Thorac Cardiovasc Surg 126:1943-1951, 2003
Viney RC, Boyer MJ, King MT, et al: Randomized controlled trial of the role of positron emission tomography in the management of stage I and II non-small-cell lung cancer. J Clin Oncol 22:2357-2362, 2004
Schirrmeister H, Hetzel M, Buck A: Staging of non-small-cell lung cancer with integrated PET and CT. N Engl J Med 349:1188-1190, 2003(Francisco Pozo-Rodríguez,)