Despite advances in cancer therapy, including gene therapy, lung cancer remains difficult to treat. The average patient with newly diagnosed lung cancer has barely a one in 10 chance of beating the disease. Each year, more than 170,000 people in the U.S. develop lung cancer and 160,000 die from it.

The quest for new or improved therapy continues, but cancer specialists also hope that earlier detection of lung cancer in a larger number of patients will help change the dismal prognosis. Stage I lung cancers can have a cure rate of up to 70% or higher, but few tumors are found at that early stage.

Chest x-rays to screen for lung cancer were common in the 1950s and 1960s. These screening programs yielded few early, treatable tumors, however, and the campaign faded. But interest in lung cancer screening programs has revived, with the focus shifting from x-rays to CT scanning. Researchers are targeting the screenings to high-risk patients, primarily smokers, rather than the general public.

In September, the National Cancer Institute announced a plan to recruit 3000 current and former smokers for a $3 million study using spiral CT.

The study at this point is not designed to show whether spiral CT scanning can save lives, said lead investigator Dr. John Gohagan, chief of the Early Detection Research Group for NCI’s Division of Cancer Prevention. In its first phase, six screening centers (Figure 1) will each recruit 500 people and randomly assign them to receive either a spiral CT scan or chest x-ray. This phase will compare respective lung cancer detection rates, determine medical follow-up for positive or inconclusive results, and track whether participants receive spiral CT scans outside the study.

Board-certified radiologists will review each CT scan and x-ray, mailing results to participants and their physicians within three weeks of the test. Patients with positive chest x-rays will receive standard follow-up care. No similar standard exists for follow-up of screening CT exams, but patients with positive or suspicious CT scans will be referred to their physician and urged to contact a specialist.

Evaluating the willingness of smokers to participate in studies is an important factor of the NCI study, especially for those who will not be receiving the CT scan. These participants could have CT scans done separately, but the cost would be at least $300 and as much as $1000. Health plans generally don’t cover CT scans for asymptomatic patients.

Other concerns include cost-effectiveness. No one has determined how much follow-up care for CT-screened patients will cost, and some experts worry that screening programs could yield a high number of false positives, as many smokers have scarring in their lungs that mimics tumors.

The NCI study is not the only one focusing on the potential of CT as a screening tool for lung cancer. For nearly a decade, New York Weill Cornell Medical Center has led the Early Lung Cancer Action Project (ELCAP). Researchers with ELCAP have shown that baseline low-dose CT screening for high-risk patients can detect cancer at earlier, potentially more curable stages than chest x-rays. ELCAP also found that annual screening of at-risk patients demonstrated better results than one-shot CT scans. The repeat screening study showed that false positives were not common and that 83% of the cancers found were in the early, most curable stages.

Advanced CT technology allowing single breath-hold scans sparked the interested in screening, said Dr. Claudia Henschke, division chief of chest imaging at Weill Medical College of Cornell.

“Without the single breath-hold scan, you had to take a slice, stop, take a slice. Each time, depending on how much the patient breathed, you may or may not have covered the same or different section of the lungs,” she said.

Overcoming Bias

Henschke and her fellow researchers realized that opinion was stacked against lung cancer screening. They reviewed all the previous studies that had recommended against screening, paying special attention to the study methods and analysis of results. They even invited one of the lead statisticians from a previous study to come and meet with them. The intensive preplanning led to the development of study models with optimistic projections.

ELCAP examined the results of 1184 repeat CT screenings performed annually on 841 high-risk individuals. Positive results were defined as newly detected, noncalcified pulmonary nodules with interim growth. The CT scans were able to find nodules as small as 2 mm.

In two of the 30 positive cases, the patients died of unrelated causes. In 12 of the remaining 28 positives, the nodules resolved on follow-up high-resolution CT (HRCT), some after antibiotic therapy. In the 16 remaining positive cases, eight had further growth, and biopsies found cancer in seven of these.

In the seven malignancies, six were non-small cell carcinomas and all were considered operable. Patients with CT-detected operable stage IA non-small cell tumors measuring less than 20 mm have a 90% chance of surviving five years and an 80% chance of cure, according to Henschke.

“If we can increase detection to something like 80% and thus increase the cure rate, we can change survival from 10% to roughly 80%,” she said.

Despite these promising results, many cancer specialists continue to be skeptical about the benefits of CT screening. Much of this skepticism is based on the Mayo Lung Project, which reported that screening tests for lung cancer, in this case x-ray, could find tumors that never become life-threatening. Specifically, the Mayo Lung Project said screening programs could lead to overdiagnosis and unnecessary biopsy or surgery. Henschke disagrees with the assumptions that screening does not benefit patients in the long run.

“The mortality rate calculation didn’t focus on the relevant period, where you could have seen the effect,” she said. “In addition, it didn’t screen long enough to start seeing the effect, and there were too many people who crossed over, who didn’t complete the screening process or who were screened even though they were in the control arm. We felt the data were inconclusive.”

International Studies

Interest in CT lung cancer screening transcends national boundaries.

A Danish study (Figure 2) ties CT screening to smoking cessation programs. Slated to start in early 2002 at Gentofte Hospital in Copenhagen, it will evaluate the practicality of setting up a low-dose CT lung cancer screening program with 10,000 regular smokers between the ages of 50 and 65. Subjects must be fit enough to undergo surgery for resectable tumors. The smokers will also be invited to participate in a concurrent smoking cessation program. Researchers plan to conduct an interim analysis after four years to examine lung cancer mortality. After five years, they will examine lung cancer mortality and other factors, such as the effectiveness of computer-aided detection (CAD). Based on ELCAP results, the Danish researchers expect to discover suspicious findings in as many as 1500 study participants.

In Japan, lung cancer is the number one cause of cancer death for men and fourth for women. Dissatisfied with the results of chest x-ray screening programs, Japan’s Anti-Lung Cancer Association conducted CT screening of 1682 high-risk individuals. A total of 36 lung cancers were detected: 24 by CT alone, four by CT and chest x-ray, four by sputum cytology alone, and four by all three modalities. Most of the cancers detected by CT were stage IA and operable.

“The five-year survival rate for the patients in all groups was 71% and 87% for patients whose cancers were detected by CT alone,” said Dr. Tomotaka Sobue of the National Cancer Center in Tokyo. “We believe CT is a valid screening method, but there were criticisms about overdiagnosis and lead time bias.”

The NCC developed a CAD system for detecting chest nodules with CT. The researchers believe its performance matches that of expert radiologists, but the study is ongoing.

Assessing CAD

Many supporters of lung cancer CT screening are hoping to improve effectiveness with CAD. U.S. researchers with the NCI believe lung cancer CT screening provides an excellent platform for assessing CAD (Figure 3).

“Lung imaging is a good physical model in that it involves the use of 3-D CAD methods that require critical software optimization for both detection and classification (benign versus malignant disease),” the NCI reported. “In addition, the detection of change in CT images over time, or change in lung nodule size, has the potential to provide either improved early cancer detection or improved classification.”

With the number of patients enrolled in CT screening studies increasing, CAD may bring several benefits. These include improved sensitivity of cancer detection, reduced errors and variation in image interpretation, increased efficiency of reading scans, improved screening efficiency due to flagging of suspect lesions, and improved remote reading.

The NCI’s proposed CAD initiative would support a consortium of centers to construct a database of spiral CT lung images. The biggest obstacle to lung cancer CAD is the lack of a process to develop consensus and standards for building and evaluating the database, according to Dr. Edward Staab at the NCI’s Biomedical Imaging Program in Bethesda, MD.

Within the imaging industry, several companies are exploring CAD possibilities for lung cancer screening, but the emphasis is on computed and digital radiography. Kodak and R2 Technologies are working to develop a commercial CR product that incorporates lung nodule detection. R2 is also working with Cornell’s lung cancer CT program by conducting comparative research with CAD for CR chest images. Deus Technologies of Rockville, MD, introduced its RapidScreen lung nodule detection tool, the first approved by the FDA, which digitizes and analyzes chest x-rays.

Multidetector CT (MDCT) may have a place in lung cancer screening programs as well. At the University of Pittsburgh, a team of researchers led by Dr. Joel Weissfeld will conduct a five-year MDCT screening study of 6000 high-risk participants aged 50 to 79. All patients have smoked 11 or more cigarettes per day for at least 25 years. Those who have quit smoking must have stopped no more than 10 years before entering the study. Participants will have a baseline screening after enrollment and follow-up screening every two years.

The Pittsburgh MDCT project is part of the NCI’s Specialized Programs of Research Excellence (SPOREs), and the study goes beyond collecting image data to include blood and DNA samples. Weissfeld and his colleagues hope to track genetic susceptibility and biochemical biomarkers in relation to lung cancer risk. MDCT will be assessed not only for its ability to detect small lung cancers, but also for its role in combined clinical, questionnaire, and laboratory data.

Advocates of MDCT for lung cancer screening maintain that the technology has several advantages, including rapid patient throughput and the use of true volume acquisition for better data visualization. The three-dimensional volume displays from MDCT scanning with either volume-rendering technique or maximum intensity projection (MIP) may prove to be more accurate than cine displays, especially when distinguishing between small nodules and vessels.

Even the MDCT and CAD studies acknowledge their debt to ELCAP for reviving interest in lung cancer screening and bringing CT into the lineup.

“Screening with CT is economically feasible,” Henschke said. “First, early-stage treatment costs less—about half—than late-stage treatment, and we all know that the potential cure rate is much higher in early-stage cancer. Resistance comes from those who have an orthodox view of how screening should be assessed.”

Figure 1.

NCI Lung Cancer CT Screening Participants

• Georgetown University Medical Center/Lombardi Cancer Research Center, Washington, DC
• Henry Ford Health System, Detroit
• University of Minnesota School of Public Health/Virginia L. Piper Cancer Institute, Minneapolis
• Washington University School of Medicine, St. Louis
• Marshfield Medical Research and Education Foundation, Marshfield, WI
• University of Alabama at Birmingham

Figure 2.

Scanning Protocol for Baseline and Annual CT Lung Cancer Screening Program in Denmark

All scans to be performed in spiral mode with the following parameters:

• 120 to 140 kV
• 20 mAs
• 1-mm single-slice collimation
• 7-mm table feed per rotation (pitch = 1.75)
• Scan direction: caudocranially (to prevent breathing artifacts) and smallest field-of-view to include outer rib margins at widest dimension of thorax
• Patients examined in supine position at suspended maximal inspiration after a three-breath hyperventilation.

Source: Gentofte Hospital, Copenhagen, Denmark

Figure 3.

NCI Goals for CAD in Spiral CT Lung Cancer Screening Programs

• Develop criteria for
• Submitting cases representing good clinical practice
• Determining reference standards for lung nodules electronically in 3-D
• Provide common research resource to medical imaging community to
• Identify promising software methods
• Stimulate development of advanced 3-D CAD methods, including temporal analysis and related image registration methods
• Accelerate research timelines
• Reduce risk for diagnostic software developed by academic facilities and/or industry
• Allow Internet access to the database by the imaging research community

Source: National Cancer Institute