CT Screening of Lung Cancer: Controversies and Challenges Abound Glenn Krinsky, MD Disclosures The utility of computed tomography (CT) screening for lung cancer remains a controversial and hotly debated topic among radiologists, pulmonary specialists, payers, and patients. Since the findings of the ELCAP (Early Lung Cancer Action Project) study were published in The Lancet in 1999, clinicians in many imaging centers have been screening patients for lung cancer with CT, even in instances in which the patients have no risk factors, such as smoking. These studies cost between $200 and $800, depending on the fee schedule for the center in which the study is performed. However, without data from a randomized clinical trial to assess the utility of screening in improving the detection and management of pulmonary malignancies, the benefits of screening patients at risk for lung cancer remain unknown; in fact, in certain geographic populations (such as the histoplasmosis belt) CT detects benign nodules in virtually all patients who have this test. The phenomenon of "questionable pulmonary nodules" detected on CT screening studies may lead to an enormous number of unnecessary diagnostic scans to further characterize or to assess the lesion for interval growth. In addition, this approach could lead to an increased number of lung biopsies and wedge resections to definitively detect whether a lesion is malignant, which will certainly escalate healthcare costs. There is an inherent critical dilemma in effective utilization of this new screening technology: If lung cancer screening detects curable stage I cancer that would progress to incurable stage III disease if not detected, then such screening may save lives. Conversely, clinicians may overdiagnose lung cancer through this technology by detecting lesions that may be neoplastic but may not ultimately result in mortality. This critical and swiftly evolving topic was addressed by researchers in this important area of radiology at 2 separate sessions of the 87th Scientific Assembly and Annual Meeting of the Radiological Society of North America (RSNA). Phantoms and Computer-Aided Diagnostic Systems Verify CT Findings Kostis and colleagues[1] used a phantom to demonstrate the importance of how different scan parameters for low-dose multidetector CT scans can affect the ability to detect calcium in small pulmonary nodules. Detection of calcification is critical because calcified nodules are almost always benign granulomas. Because attenuation measurements are inversely related to slice thickness and pitch, thicker slices and higher pitch reduce the sensitivity of calcium detection within a nodule. Ko and researchers[2] evaluated the use of wavelet compression algorithms (lossy uncompressed, 10:1, 20:1, and 50:1) in the detection of pulmonary nodules with low-dose multidetector CT scanning. Sensitivity for detection of pulmonary nodules was the same at 10:1 compared with the uncompressed acquisition but decreased significantly at 20:1 and 50:1. Increasing levels of compression did not change diagnostic specificity. Low-dose CT and a computer-aided diagnostic (CAD) technology can improve the diagnostic accuracy of screening CT. Li and researchers[3] reported on a study involving approximately 18,000 patients who underwent screening with 10-mm collimation low-dose CT. Eighty-two cases of lung cancer were diagnosed during the prevalence (first) screen, and 42 cases were detected during the 2-year incidence (follow-up) screen. Retrospective review of the prevalence screen studies detected 30 "missed" cases, which were located in the lung periphery. Of these, 20 lesions were thought to be detection errors and 10 cases were ascribed to interpretation errors. The primary reasons noted for detection errors were small nodule size and density, overlapping pulmonary vessels and fissures, and lack of attention to imaging of the pulmonary hilum. Interpretation errors were caused by confusion created by findings referable to underlying inflammatory lung disease. CAD studies detected 78% of the missed lesions and proved to be a promising tool to decrease the number of missed cases of cancer due to human error. Volumetric Growth Index, Utility of the Maximum-Intensity-Projection Algorithms, and Thin-Slice Acquisitions Reeves and colleagues[4] have developed and tested a volumetric growth index (VGI) technique for detection of pulmonary nodules. This tool, which uses a 3-dimensional computer-aided index, shows promise in distinguishing malignant from nonmalignant pulmonary nodules in less than 6 months, even in small lesions. Ko [2] also noted during the question-and-answer session following this presentation that using a 3-dimensional algebraic method can improve nodule quantification by decreasing partial volume effects. It is clear from these data that the volumetric approach will be critical for accurate measurement of the growth of small lung nodules in the future, because traditional bidimensional orthogonal measurements are often inaccurate in this regard, especially in cases in which the lesion is not perfectly round. Brown and coworkers[5] demonstrated that baseline image data from a prevalence screen can be used to create a patient-specific model of anatomy suitable for segmentation and tracking of nodules on subsequent CT scans. By automatically relocalizing nodules, this technique may reduce the number of false-positive lesions detected on subsequent examinations. It can also track multiple nodules over time, which facilitates accurate recognition of nodule growth. Two presentations during this session evaluated use of the maximum-intensity-projection (MIP) algorithm to improve the sensitivity of multidetector CT for detection of small pulmonary nodules. Gruden and Tigges[6] retrospectively reviewed 112 nodules < 1 cm in size in a group of 25 patients. On routine axial scans, the readers missed 30% to 70% of central lesions and 8% to 40% of peripheral nodules. The addition of MIP images reduced error rates by 40% to 55% for all observers. Miss rates varied significantly and correlated to reader experience.Fischbach and coworkers[7] related research that used low-dose multidetector CT scans at 1-mm collimation and scans reconstructed at intervals of 2.5 and 5 mm to determine whether thinner slices improve detection of pulmonary nodules. In this study, 14 of 42 nodules were detected only on the 1-mm raw data, and most of the lesions detected were < 5 mm. In addition, thinner slices also enhanced diagnostic specificity because of their ability to differentiate between small vessels and nodules. This presentation highlighted the importance of careful analysis of raw data in establishing an accurate imaging diagnosis of pulmonary nodules. Overdiagnosis of Lung Cancer: Fact or Fiction? Some clinicians and radiologists believe that lung cancer is being overdiagnosed by imaging studies. This issue was evaluated by Yankelevitz and colleagues[8] through a retrospective study evaluating data from the Mayo Clinic and Memorial Sloan-Kettering lung cancer screening projects performed in the 1970s with chest radiography. This group focused only on stage I lung cancer, assuming that that is the only subset of patients in which overdiagnosis would be of concern. The median size of stage 1 tumors were 2 and 2.2 cm and the doubling times were 130 and 150 days, respectively. The relatively rapid growth rate of lesions this size was used to argue that cases of lung cancer detected in a screening population are indeed lethal neoplasms and not a disease of technology (overdiagnosis). In a presentation that addressed the same topic, Kakinuma and investigators[9] evaluated the natural history of small (< 15 mm) lung lesions over the course of an 8-year lung cancer screening program in Japan using 6-month incidence studies. Ground-glass opacity nodules were invariably bronchoalveolar carcinoma, while the solid nodules were invasive carcinomas. Mean doubling time for the bronchoalveolar carcinomas was 1091 days (range, 221 to 2674) and 139 days (range, 48 to 312) for solid tumors. The data from the group suggest that small, incidental lesions detected at screening CT can grow rapidly and are therefore potentially lethal.Lee and investigators[10] screened 457 asymptomatic high-risk patients (smoking > 20 pack-years) with low-dose, single-slice, helical CT with 7-mm collimation. The prevalence screen detected 286 nodules in 143 (31%) patients. Only one case of lung cancer was detected, which was also seen on chest radiography. Given the high prevalence of granulomatous disease in Korea, the authors advocate tailoring screening and follow-up schemes to the regional environment.Garg and colleagues[11] tested the hypothesis that people might not agree to participate in randomized, controlled trials for lung cancer screening. Their preliminary results in 112 persons indicate a high degree of willingness -- especially in persons at higher risk -- to be randomly assigned to have or not to have CT screening. Yankelevitz and colleagues[12] demonstrated that cytologic studies from fine-needle aspiration of lesions as small as 4 mm detected with screening CT can provide a definitive diagnosis in most cases. In this study, 12 of 65 patients had atypical bronchiolar proliferation, also known as atypical adenomatous hyperplasia. It will be interesting to see the imaging and clinical follow-up of this group of patients, who have ground-glass opacity nodules due to atypical bronchiolar proliferation, to determine if this condition is a precursor to bronchoalveolar carcinoma. MEDSCAPE 4/02 Lung Scans Arch Intern Med, 2/06 Remember we are NOT Doctors and have NO medical training. This site is like an Encylopedia - there are many pages, many links on many topics. Support our work with any size DONATION - see left side of any page - for how to donate. You can help raise awareness of CAM.