On June 24, 2013, in Bethesda, Maryland—at the second joint meeting of the National Cancer Institute (NCI) board of scientific advisors and the National Cancer Advisory Board—William C. Black, MD, a Dartmouth radiologist, copresented “National Lung Screening Trial1 (NLST) Subset Analysis,” a cost-effectiveness analysis of low-dose CT (LDCT) screening for lung cancer. His data showed that LDCT not only was a cost-effective method of screening for lung cancer, but was more cost effective than some other screening exams (such as screening mammography).
“Very shortly thereafter, on July 30, the United States Preventive Services Task Force (USPSTF) posted its revised draft recommendations with a systematic review to update its 2004 guidelines and invited public comment through the end of August,” Ella Kazerooni, MD, related to those gathered at a special session, “Lung Cancer Screening: How I Do It,” on December 1, 2013, at the annual meeting of the RSNA in Chicago, Illinois. “The draft new recommendation was to give lung-cancer screening with LDCT a grade B recommendation for high-risk individuals with a smoking history,” she explains.
During the special session, Kazerooni gave a current-data summary and update. Reginald F. Munden, MD, DMD, MBA, described how to start a screening program; Thomas E. Hartman, MD, provided management strategies for screening-detected nodules; and David F. Yankelevitz, MD, contributed pearls for the biopsy of screening-detected nodules. Caroline Chiles, MD, offered guidance for the management of incidental findings.
The final B grade recommendation from USPSTF, announced later in December, set in motion a series of events that has put lung-cancer screening in the high-risk population on the fast track to reimbursement. While it might appear, to casual observers, to be a sudden development, the ACR®; the NCI; and many other organizations, radiologists, radiology departments, and practices have invested a great deal of time, thought, and energy in determining what an evidence-based program for LDCT lung-cancer screening should look like.
Summary and Update
The USPSTF draft recommendation on lung-cancer screening that Kazerooni flags as a milestone calls for annual LDCT from 55 to 79 years of age for individuals with a history of heavy smoking (if they currently smoke or have stopped within the past 15 years) who have no significant pulmonary disease. As mandated by the Patient Protection and Affordable Care Act (PPACA), a recommendation of B or above from the USPSTF means that private insurers must cover the procedure (with no deductible). While the PPACA does not similarly obligate Medicare, CMS initiated a National Coverage Analysis on February 10, 2014.
Within organized radiology, there has been a great deal of activity aimed at developing standards and resources for lung-cancer screening, Kazerooni (who chairs the ACR Committee on Lung Cancer Screening) says. “The ACR is working on many fronts to provide information to people who seek lung-cancer screening or seek to provide the service,” she reports.
A practice guideline for the performance and interpretation of CT screening for lung cancer is nearing its final review and could be approved at the ACR Annual Meeting and Chapter Leadership Conference (April 26–30, 2014). Appropriateness criteria covering which patients should be screened also are under development.
In progress, as well, is the development of LuRADS, a lung reporting and data system structured as the successful ACR BI-RADS® is (for mammography reporting). The system will feature a number-based rating method for categorizing screening findings and the management strategies associated with those categories.
The college also is devising a variant of facility accreditation, as well as patient- and radiologist-education materials. “If you are going to screen,” Kazerooni asks, “what are the important things you need to know to interpret, perform, and report?”
Radiologists interested in participating in lung-cancer screening are advised to visit the website (www.lungcanceralliance.org) of the Lung Cancer Alliance, the only national nonprofit organization that exists solely to provide support and advocacy for people living with lung cancer. “It has really influenced the landscape in moving forward the agenda to get lung-cancer screening reimbursed,” Kazerooni says.
She is optimistic that lung-cancer screening will become public policy in the United States. “With the cost-effectiveness data and the movement from the USPSTF, we are in a very stable position to move forward,” she notes. “We should look forward to this being rolled out, in populations at risk, as the first (and only) way shown to reduce lung-cancer mortality.”
Setting Up a Program
Take it from one who knows: It is imperative that your program for lung-cancer screening be multidisciplinary. Munden, an NLST principal investigator who set up a screening program at the University of Texas MD Anderson Cancer Center in Houston in 2011 (and is now chair of the department of radiology at Houston Methodist Hospital and Houston Methodist Research Institute), speaks from practical experience.
“If you don’t have the right team, patients will fall through the cracks, and the results of your screening, perhaps, will not be as good as they should be,” he says. Radiologists were instrumental in proving the efficacy and cost effectiveness of LDCT lung-cancer screening, but they do not control the screening process from end to end. Primary-care physicians, surgeons, pulmonologists, health-care providers, and others who might refer and follow patients (such as oncologists or radiation therapists) should all be included.
It is not necessary to reinvent the wheel, as many radiology providers offer mammography services; many elements (such as the services of a medical physicist who verifies that equipment is operating properly) can be adapted for lung-cancer screening. “As the ACR guidelines come down, I am sure we will have as many (if not more) guidelines, standards, and regulations as we have in mammography,” Munden notes.
Having a coordinator (sometimes called a patient navigator) is imperative. The coordinator serves as the point person for both patients and referring physicians, keeping track of where patients are in the program and when they are due to return. “This person is the unifying glue, holding everything together,” Munden says.
One of the first decisions that program operators must make is which patients will be eligible to be screened. The NLST inclusion criteria call for patients to be 55 to 74 years old, to have a smoking history of 30 or more pack years, to be current smokers or to have stopped smoking within the past 15 years, to be able to lie on their backs with their arms raised over their heads, and to have signed a consent form. Exclusion criteria are extensive. Program operators can consult the National Comprehensive Cancer Network (NCCN), USPSTF, and professional-society recommendations for guidance on program entry.
Making It Successful
Once criteria have been selected, telephone operators must be trained to qualify patients and transfer calls effectively, asking questions to discover the patient’s age and to calculate pack years. “We’ve had success,” Munden says. “We haven’t screened people in our program who shouldn’t be screened.”
Munden, however, advocates the ability of patients to self-refer, as they can for mammography, because it removes an obstacle to entry. If there is someone at the hospital willing to take responsibility for that patient, and if legal issues (and state–medical-board politics) can be addressed satisfactorily, that might be an option.
He also recommends sending reports to all patients. “For a number of reasons, a report sent to a physician’s office may not be acted upon in a timely manner, which can result in unfortunate outcomes we want to avoid. If the report goes to the patient, this is much less likely to occur,” he says.
The current self-paid price for screening ranges from $300 to $400, Munden says, and some private payors are reimbursing for the study. The USPSTF grade B recommendation ensures that reimbursement by private payors is forthcoming, but the amount is up for discussion. “I wouldn’t count on $400,” he says. “I suspect it will be in the range of $200 to $250, and it may be as low as $100.”
Many in the screening community agree that a smoking-cessation program should be included—because success rates are better when those programs are tied to lung-cancer screening. The screening program’s coordinator meets with every patient, provides a smoking-cessation brochure, and details the options available.
Generating—and maintaining—interest, particularly in annual follow-up, is not easy, Munden says. Billboards and advertisements can be used, but showing up at health fairs and other community events seems more effective than other direct-marketing initiatives. Consider partnering with local industry in lung-cancer screening, which could be bundled with other screening programs.
After your program has been established, look for patient advocates: “Anyone who is a survivor because of your efforts is vital to your program,” Munden believes. “These survivors can add insight, and they are instant marketing for screening: They get the word out about the success of what you are doing.”
In conclusion, Munden emphasizes the importance of standardization, structured reporting, and participating in research to improve the screening process. At a minimum, it is vital to quality-assurance (QA) studies that radiologists follow some percentage of patients with positive findings.
“The biggest negative in the NLST is the high false-positive rate of 96%,” Munden says. “That’s huge, but through standardization, QA initiatives, and research, we can dramatically improve that rate.”
Management of Nodules
Reading a chest CT will not be the challenge for radiologists: Knowing when to act on results is where the learning curve lies. The first question that a radiologist must ask in management of a screening-detected nodular opacity, Hartman says, is whether it needs to be managed. Calcified granulomas, hamartomas, and perifissural nodules do not. Hartman, an NLST coinvestigator, subspecializes in chest radiology at the Mayo Clinic (Rochester, Minnesota).
Perifissural nodules are solid nodules that are lie along a fissure. They are triangular in shape, with the broadest base along the fissure, and in the setting of lung-cancer screening have not been shown to be malignant (regardless of size). Showing an image of a perifissural nodule 6 mm in diameter, Hartman says, “If this were the only nodule you saw, this would be a negative finding, and you would not have to do anything but bring this person back for an annual screening.”
When they lie along the minor (horizontal) fissure, it is more difficult to discern whether they are perifissural nodules, but reconstructions with sagittal and coronal dimensions can be done in the background on many PACS and can confirm the perifissural location, Hartman notes. When a screening-detected nodule is none of the above, there are ways to limit the number of nodules worked up—and therefore, the number of false positives. The limiting factors include size, margins, and attenuation; computer-aided detection also can be helpful.
Nodule Size—Using the NLST protocol (the only protocol proven effective in the reduction of lung cancer through screening), a screening exam is positive if a noncalcified nodule is found that is 4 mm or larger in its greatest transverse dimension. The presence of a noncalcified nodule of less than 4 mm is considered a negative screening, and the patient returns for annual screening.
Nodules of 4 to 10 mm call for six-month follow-up with LDCT. If they are growing at follow-up, they are candidates for more aggressive work-up, which could include PET/CT imaging or biopsy. Nodules that are 10 mm or larger at the time of initial detection also undergo more aggressive work-up with PET/CT or biopsy evaluation. Hartman explains that the NLST guidelines do not really have any other specific criteria for the margins or attenuation of the nodule, but they do allow modification of the recommendations based on those characteristics.
While only the NLST criteria have been evaluated in a controlled study, there are likely to be changes as the discipline matures, including changes in the nodule-size guidelines. Hartman cites a study2 based on data from the Early Lung Cancer Action Project Trial, which used a 5-mm threshold for nodule size—yielding 3,400 positive exams, but just 119 lung cancers. By retrospectively moving the threshold up to 9 mm, the number of positive exams dropped to 838, with 111 cancers detected.
“By changing the nodule-size threshold, the investigators were able to decrease the false positives from 16% to 4%,” Hartman notes. “Although there were eight cancers that would have been missed, initially, with a higher nodule-size threshold, all of those cancers were at stage I, and it is debatable whether those cancers would have had an impact on patient mortality.” Based on work suggesting that the nodule-size threshold can be increased without negatively affecting mortality, the NCCN recommendations for follow-up differ slightly on nodule size from those of the NLST.
Nodule Attenuation—The ACR and the Society of Thoracic Radiology (STR) are working to finalize guidelines for screening, but the ACR has recommended the NCCN guidelines as appropriate to follow in the interim. The NCCN identifies three separate nodule types: solid, partly solid, and ground glass.
For both solid and partly solid nodules, if a nodule is smaller than 6 mm, the patient would go back to annual screening. In the 6–8-mm category, LDCT at three months is recommended—and if there is no growth, again at six months. If there is growth, more aggressive work-up (including possible PET/CT or biopsy) should be considered. For any nodule larger than 8 mm, more aggressive work-up should be considered at the time of detection.
Nodule attenuation also can be a modifier for how aggressively a nodule is worked up, Hartman notes. Semisolid nodules that persist on screening follow-up have been shown to be malignant more often than ground-glass and solid nodules.
For ground-glass nodules of less than 5 mm, the NCCN recommends annual screening; if ground-glass nodules measure 5–10 mm, LDCT at six months is recommended. If the nodule is stable, the patient continues with annual screening. If it’s growing, more aggressive work-up should be considered. For ground-glass nodules of 10 mm or more, the recommended follow-up interval is three to six months. If the nodule is stable and still persistent after three to six months, more aggressive work-up should be considered because persistent ground-glass nodules of that size have a higher likelihood of malignancy.
Multiple ground-glass nodules are a subarea in the NCCN guidelines, based on the size of the largest nodule: If all are greater than 5 mm (but none are dominant), follow-up CT in six months is recommended; if the nodules are stable at that time, the patient returns to annual screening. If all nodules are larger than 5 mm, and there is one significantly larger than the others (or with a partly solid component), the recommended follow-up is CT at three to six months, to see whether it has persisted. If so, more aggressive work-up of that nodule should be considered.
Nodule Margins—The tendency to manage nodules based on whether they are smooth, lobulated, regular, or spiculated goes back to the early 1990s, Hartman says. “Given the higher likelihood of malignancy in spiculated nodules, they are something we are going to want to be more in tune with aggressively working up, versus a smooth nodule,” he adds.
Computer-aided Detection—Hartman says that computer-aided detection can be helpful in assessing the morphology of the lesion, as well as in volumetric analysis. Morphological analysis of lung nodules remains a bit tricky, as there have been conflicting results concerning its utility.
“Where it really has shown significant benefit—or, at least, promise—is in volumetric analysis,” Hartman says. “It’s been shown to be more accurate and reproducible in predicting nodule malignancy.”
Computer-aided detection with volumetric analysis was used in the Dutch-Belgian Randomized Lung Cancer Screening (NELSON) Trial to determine whether nodules 4–10 mm in size should indicate a positive screening. Nodules with volume-doubling times of less than 400 days are considered positive; those with longer doubling times are negative, and the patient goes back to annual screening.
Hartman identifies a number of concerns that need to be addressed before these tools can become part of a screening program. They are nodule size, segmentation challenges, radiation-dose variation, variations introduced by differences in inspiratory level, and variation between algorithms.
Smaller nodules, for instance, show an increased error in the measurement of doubling volume, he says. Segmentation is more difficult to determine when a nodule has contact with adjacent vessels and pleura, and the volume-assessment error for pleural nodules is higher than for other types of nodules. The mean absolute error for volume measurements is smaller for standard-dose (versus LDCT) images, making dose an additional factor.
When comparing studies with different inspiration efforts, radiologists need to be circumspect about volume measurements, Hartman says. There are significant differences in mean nodule volume between studies done on inspiration and on expiration.
Different software packages use different algorithms to calculate volume (and they produce different results), so nodule volumes should not be compared between different software packages. The idea of nodule mass—nodule volume plus mean attenuation—shows promise, especially in evaluating ground-glass nodules that increase in attenuation or become semisolid.
Biomarkers are another area of active study with the potential to take nodule management beyond size, either by enriching the pretest screening pool or by determining whether a nodule should be worked up more aggressively. “At the end of the day, the goal is to optimize the identification of malignant nodules,” Hartman concludes. With nodule size being the only proven way to stratify and manage patients, he says, developing standardized protocols for both screening and nodule management is imperative.
“In the future, we will have some additional tools,” Hartman says. “For now, though, nodule size is the primary criterion, and the guidelines from the ACR/STR that will be released soon should allow us to standardize our practices.”
Biopsy of Screening-detected Modules
In general, when a lesion requires more aggressive evaluation, the choices are repeat LDCT, PET/CT, or biopsy. Yankelevitz, director of the lung-biopsy service at Mount Sinai Hospital (New York, New York), is likely to have performed more fine-needle aspirations of lung nodules than many radiologists have. Based on his involvement in the International Early Lung Cancer Action Program, he shares indications and techniques for biopsy and pearls for improving specimen evaluation (which has greater-than-ever implications for treatment).
“The field of lung cancer is now changing more rapidly than at any time, and we are already finding targeted agents not just for adenocarcinoma, but for squamous-cell carcinoma,” he says. In the context of screening, however, the goal is to discover whether a screening-detected nodule is malignant. A nodule’s growth rate, size, or some combination of these factors leads to biopsy. The target has gotten smaller since Yankelevitz began his screening program, in 1992, but the guidelines for management are in flux.
Growth Rate—Growth assessment is very appealing and actually grew out of screening; Yankelevitz says, “This is now one of the dominant methods of evaluating these nodules.” There are no definite answers yet to the question of how much growth represents real change, but researchers are coming to consensus around 30% at a 10-mm size and 70% to 80% for smaller nodules. “Clearly, the smaller the nodule, the more proportional change you need,” he says. “Scanners are changing all the time. This is going to be a continually moving area of evaluation.”
Calling this the new reality of screening, Yankelevitz shares a series of images acquired over time, beginning with a barely visible nonsolid area on a chest CT image. A year later, a tiny solid component has developed; later, there is a 5-mm partly solid nodule with a 2-mm solid component.
“I don’t know what to do with these, and I’ve been doing this since 1992,” he says. “Would you keep sitting on this? It’s small.” He was asked to biopsy the lesion, which turned out to be adenocarcinoma.
Thoracoscopy—Yankelevitz used to believe that there was no role for bronchoscopy in screening, but has softened his stance a bit. He still maintains that there is no role for thoracotomy in biopsy, but going straight to thoracoscopy would save a step, if a lesion turns out to be malignant. Knowing with an 80% or 90% certainty that a lesion is malignant will, one day, help guide follow-up.
Likewise, having confidence that a negative biopsy is truly negative is another hurdle. If you have documented the needle in place, you can be confident that over 90% of the time, the lesion is benign, Yankelevitz says. Hitting that mark, however, requires skill.
Bite Like a Mosquito
“Sometimes, you put that needle in, and it goes right where you want it: It’s like magic,” Yankelevitz says, “but here are the facts. You put a needle in place: How well do you have to line up that needle? I used to do this for 1-cm nodules, and now, I’m doing it for 3-mm nodules. How well do you have to position the angle of that needle?” Yankelevitz says that given the reality of a breathing patient who is likely to be uncomfortable on your table, it’s impossible to line up the needle within a single degree of its mark.
Having spent a good deal of time thinking about how to solve this problem, Yankelevitz found his answer in the mosquito, which he calls the world’s best performer of biopsies: “All of you have experienced these types of biopsies,” he says. “How do they do it?” Yankelevitz discovered that when a mosquito inserts its proboscis, it goes in straight, but the appendage has sensors that enable the insect to determine where a blood vessel can be found. When mosquitoes detect blood vessels, they bend their proboscides.
The technique can be replicated using a beveled needle. When you press on a bevel-tipped needle, the forces applied are asymmetrical, and the needle bends away from the bevel. This technique enables Yankelevitz to retract the needle partially, turn the bevel in the opposite direction, torque the needle, pull the skin, and correct course without having to repuncture the pleura. “Once you can do this (and it takes a lot of practice), you can biopsy a nodule of any size—anywhere,” he says.
An important member of a multidisciplinary program for lung-cancer screening is the cytologist, Yankelevitz emphasizes. He says, “Develop a close relationship with your cytologists. If you are going to do this, they have to be members of the team.” He does 20 biopsies a week—too many to expect a cytologist to examine every specimen on the spot. “Try to get your organizations to dedicate cytotechnologists to work with you,” he suggests. “They can be outstanding.”
Review the cell blocks, and have them audited if you are consistently unhappy with the yield. “There are many ways to prepare specimens,” Yankelevitz says. Work with the cytologists to help them become comfortable with assessing small numbers of cells— perhaps only one, if it clearly is cancer. He explains, “You don’t want to keep poking these difficult-to-biopsy lesions.”
There are many potential incidental findings in the neck, thorax, and upper abdomen, but Chiles focuses on abnormalities that are significant contributors to morbidity and mortality in the population being screened for lung cancer. Chiles is professor of radiology in the Comprehensive Cancer Center at Wake Forest Baptist Medical Center (Winston-Salem, North Carolina) and a principal investigator for the NLST.
When considering an incidental finding, it helps to weigh the potential benefit to the patient of making this observation versus the additional cost of the diagnostic work-up, the creation of unnecessary patient anxiety, or—even worse—unnecessary injury due to unnecessary work-up. “It’s easy to get blinded by thinking that lung cancer is the most important cause of mortality in the older smoking population,” she notes. “Look at the other causes of mortality in the NLST.”
The three main causes of mortality in the NLST screening population (in addition to lung cancer) were cardiovascular disease (very significant), respiratory disease, and other cancers. Chiles is not suggesting that radiologists should not report, for instance, the aortic aneurysms that they see in the course of reviewing screening images, but she does recommend focusing on the big three.
Cardiovascular Findings—“Even though we are doing these unenhanced LDCT scans, and they are not gated, we can still see calcification,” Chiles notes. “In fact, we can see it pretty well, given this low-dose technique.” Jacobs et al3 looked at a case cohort (within the NELSON Trial) of 150 patients who experienced either cardiac deaths or cardiac events, excluding subjects who had a known history of cardiovascular disease. They calculated Agatston scores for coronary calcifications and divided them into four quartiles of risk: 0 (no calcification), 1 to 100, 101 to 1,000, and more than 1,000. The researchers found that an Agatston score of more than 1,000 was an independent predictor of all-cause death, as well as of fatal and nonfatal coronary events.
Within the NLST, five cardiothoracic radiologists reviewed about 1,600 LDCT scans from three separate groups: The 210 people in group 1 died of coronary heart disease, the 314 people in group 2 died of causes other than coronary heart disease, and the patients in group 3 (controls) were alive at the time of the trial. Using the same four risk categories as the NELSON Trial, the NLST radiologists devised two qualitative analyses to replace the more time-consuming Agatston scoring method: An overall gestalt was noted; then, each coronary-artery segment was assigned a score of 1, 2, or 3 (associated with a reference image), for a total score of 1 to 30.
“We know that zero calcification is a good negative predictor in a patient who is a nonsmoker,” Chiles notes, “but look at this: 12% of patients who died of coronary heart disease had no calcification on their baseline screenings. We know that the absence of calcification in a nonsmoker predicts excellent survival, but the absence of calcification in a smoker is not as favorable. They behave more like nonsmokers with mild calcification, so zero calcification is not quite as good a negative predictor as in a nonsmoker.”
The NLST researchers found that overall gestalt performed similarly to the Agatston-score method in comparing hazard in patients with and without calcification, and they concluded that the gestalt method was adequate for assessing risk. “The value of this is that patients who are aware that their CT scans show coronary-artery calcification, particularly if they have a high scores, are more likely to be treated with a statin,” Chiles says. “It changes patient management. Patients also are more likely to adhere to statin therapy and to make healthy lifestyle choices, such as changes in diet and exercise.”
COPD—The fourth leading cause of death worldwide, COPD, remains underdiagnosed, even though it has a high prevalence in the screening population. “It is infrequently self-reported,” Chiles says. “Patients just don’t know they have COPD, but we know that if they have COPD diagnoses, it can improve patient management: There will be fewer COPD exacerbations and fewer trips to the emergency department. I do think it is worthwhile to evaluate this in the screening population.”
Keep in mind that COPD is an independent risk factor for lung cancer, she adds, predicting that the presence of COPD will ultimately play a role in patient-management strategies. If patients are seen to have COPD on initial screening, the fact that they are known to be at higher risk for lung cancer might influence decisions concerning how frequently to follow them and when to do the next screening.
There are two primary types of COPD: emphysema-predominant COPD and airway-predominant COPD. They can be measured qualitatively, using a gestalt method similar to that applied to coronary calcification and categories of zero, mild, moderate, or severe obstruction. There also are software techniques for the quantitative assessment of emphysema and for airway measurement.
A study by Han et al4 established a relationship among COPD exacerbations, emphysema severity, and airway-wall thickness. At lower levels of emphysema, airway-wall thickness became the predominant factor.
Other Cancers—Chiles cites work exploring the clinical implications and added costs of follow-up for suspected lesions seen in organs other than the lung (most commonly, the thyroid, kidneys, and adrenals), noting that 12% of subjects can exhibit such incidental findings. Another study,5 based on incidental findings in the NELSON Trial’s population, found what its investigators called a neglectable benefit in searching for incidental findings using LDCT.
In conclusion, Chiles recommends that radiologists report coronary-artery calcification found on LDCT as none, mild, moderate, or heavy. She also urges that the presence and severity of emphysema be reported. As for other cancers, the third significant cause of death among the population undergoing lung-cancer screening, Chiles concludes that there is limited opportunity to detect extrapulmonary malignancy. She says, “There is insufficient evidence, in the lung-cancer–screening population, to recommend follow-up of low-attenuation lesions in the thyroid, kidneys, or liver.”
- Aberle DR, Adams AM, Berg CD, et al. Reduced lung-cancer mortality with low-dose computed tomographic screening. N Engl J Med. 2011;365(5):395-409.
- Henschke CI, Yip R, Yankelevitz DF, Smith JP. Definition of a positive test result in computed tomography screening for lung cancer: a cohort study. Ann Intern Med. 2013;158(4):246-252.
- Jacobs PC, Gondrie MJ, van der Graaf Y, et al. Coronary artery calcium can predict all-cause mortality and cardiovascular events on low-dose CT screening for lung cancer. AJR Am J Roentgenol. 2012;198(3):505-511.
- Han MK, Kazerooni EA, Lynch DA, et al. Chronic obstructive pulmonary disease exacerbations in the COPDGene study: associated radiologic phenotypes. Radiology. 2011;261(1):274-282.
- van de Wiel JC, Wang Y, Xu DM, et al. Neglectable benefit of searching for incidental findings in the Dutch-Belgian lung cancer screening trial (NELSON) using low-dose multidetector CT. Eur Radiol. 2007;17(6):1474-1482.