CT Lung Cancer Screening Yields High False-Positive Rate

ROCHESTER, Minnesota—In a Mayo Clinic study of low-dose helical CT screening for lung cancer, nearly 70% of the study participants had one or more suspicious lung nodules, but only 1.4% of all nodules proved to be malignant. The other 98.6% were benign "and therefore were false-positive findings," said lead investigator Stephen J. Swensen, MD, professor of radiology. The results, he said, offer reasons for optimism as well as reasons for doubt that CT screening for lung cancer will ultimately save lives by reducing disease-specific mortality.

The preliminary results of the ongoing phase II study included 1,520 current or former smokers age 50 years and older. Study participants had to have at least 20 pack-years of smoking; those who no longer smoked had to have quit within the previous decade. All were asymptomatic for lung cancer.

At the end of 3 years of screening, the researchers are following more than 2,800 uncalcified, radiologically indeterminate lung nodules. To date, eight participants have had surgery for removal of a benign nodule, accounting for about 20% of all operations in the series (Am J Roentgenology 179:833-836, 2002).

Recent data from other studies of CT screening for lung cancer show that about half of all nodules removed are benign, he added. Such false-positive rates, Dr. Swensen observed, "would be clearly unacceptable for a mass screening endeavor." The cost alone of the unnecessary operations would make such screening unacceptable, but of greater concern, he said, "are the morbidity and the 3.8% mortality seen with wedge resections of lung nodules in community hospitals in the United States."

Reason for Optimism

In the Mayo study, 41 lung cancers have been found to date, of which 59% were stage IA. In contrast, Dr. Swensen noted, the current rate of stage IA cancers at diagnosis in clinical practice is 15% to 20%. "Survival of patients with stage I lung cancers is 60% to 70%," he noted. Thus, "there is reason for hope," he said, that CT screening could prove cost-effective in preventing lung cancer deaths.

James L. Mulshine, MD, head, Experimental Intervention Section, Cell and Cancer Biology Branch, Center for Cancer Research, National Cancer Institute, in an interview with ONI, said that he appreciated Dr. Swensen’s discussion of the problems encountered with CT screening for lung cancer. "Yet at the end of the day," he said, "they did actually do quite well in finding a very high frequency of stage I, potentially curable cancers. But it just was an enormous effort to get there."

To David Yankelevitz, MD, professor of radiology, Weill Medical College, Cornell University, New York, the Mayo Clinic results show that "CT screening leads to diagnosis of smaller, earlier cancers, and, as we know, treatment of early cancer is more effective than treatment of late cancer. I think this is becoming more and more irrefutable, and I think this article serves to reinforce that."

Dr. Yankelevitz did not find the rate of false positives reported particularly meaningful. "We need to learn what to do with the findings," he said in an interview. "We’re not creating the findings. These nodules exist. We just need to learn as scientists and physicians how to manage them appropriately."

Dr. Mulshine concurred with a suggestion by Dr. Swensen that the solution to the false-positive problem may be serial imaging and volumetrically determined growth rates to identify which nodules require further evaluation, as is done at Cornell, and that this approach needs to be validated. (See ONI December 2002, page 19.)

"I think the Cornell method of initial imaging and then follow-up imaging to look for growth is a very promising approach to CT screening for lung cancer. It may be more generally available to community radiologists in the next few years, but it remains, at present, a research tool," Dr. Mulshine said.

The ultimate goal is development of screening strategies that can be implemented in communities across the nation, not just in specialized settings, he said. Software-driven image processing, he suggested, may be helpful in achieving that goal, noting that "the precision of these instruments for these very fine discriminations is just so much more powerful than a human being."

Dr. Swensen concluded: "The bottom line is that we are unequivocally required as scientists and health professionals to thoroughly study this new ‘beast’ before jumping into what might be a quagmire. A randomized, controlled trial is the best way to address the controversy and probably the only way that third-party payers will ever agree to pay for this screening—if a trial shows positive results."

The US National Cancer Institute has begun such a trial (the National Lung Screening Trial) comparing the efficacy of spiral CT scans and chest x-rays in reducing lung cancer mortality (see ONI November 2002, page 12).

Dr. Yankelevitz, however, questioned reliance on randomized controlled trials to study the benefits of CT screening. "Randomized controlled trials are useful to evaluate interventions," he said. "They are an inappropriate paradigm to use in the study of screening tests. This has led us to the enormous failure of mammography screening to show a benefit. I think this is an absolute tragedy, that after so many years and so many trials, mammography’s benefit is still unknown."

Treating Preclinical Disease

Dr. Mulshine noted that CT screening can detect preclinical lung disease, but there is no standard treatment for such disease. He envisions that less invasive methods of managing small lesions will evolve along with the ability of CT screening to detect such lesions.

A possible new approach could be the use of less invasive surgical procedures, such as video-assisted thoracic surgery (VAT), that allow patients to recover more quickly from the operation. Another approach would be the use of tailored radiation therapy to target small lesions. Finally, new targeted medical treatments such as aerosolized drug delivery may become possible in the future.

If such less invasive approaches prove successful, Dr. Mulshine said, "you completely change the equation as to the cost-effectiveness and the potential overall benefit of CT screening."

The use of CT screening for lung cancer has the potential to spawn medical malpractice suits, according to another article in the same issue of the American Journal of Roentgenology (page 837). In this article, Leonard Berlin, MD, Department of Radiology, Rush North Shore Medical Center, Skokia, and Rush Medical College, reported one such action involving failure to diagnose lung cancer.

"Patients who for any reason come to expect that CT screening will without exception detect any early disease that they may harbor and that that disease will be cured if discovered in its earliest stage may well respond with a malpractice lawsuit if those expectations are not met," Dr. Leonard wrote.

The malpractice risk, Dr. Mulshine said, is an important issue as are those raised by the Mayo Clinic report. "This field is developing very quickly, has enormous potential for good, and potential for harm," he said. What is needed, he stressed, is radiologists, oncologists, pathologists and thoracic surgeons working together so that lung cancer screening can "evolve to a much more adaptive, more effective approach that will have to be validated with clinical trials."