Divergent research on CT lung screening sparks more debate, fewer answers

2007 04 18 15 40 07 706

There is no doubt in the healthcare community that multislice CT finds more potentially cancerous lung nodules than chest x-ray. And it stands to reason that when any form of cancer is discovered in its early stages, the chance of a patient's survival increases greatly.

The final word on CT's efficacy in reducing mortality among patients with suspicious lung nodules, however, has become considerably less conclusive and more divergent with the release of two recent research reports.

Data from the International Early Lung Cancer Action Program (I-ELCAP), released in October 2006, found that CT lung screening of smokers and former smokers can detect curable cancers, suggesting that survival rates improve markedly when the tumors are resected. In the study, 92% of participants who had stage I cancers resected were alive five years later, while patients who were not treated died from the cancer.

The findings were published in the New England Journal of Medicine by Dr. Claudia Henschke and her colleagues at Weill Medical College of Cornell University in New York City (New England Journal of Medicine, October 26, 2006, Vol. 355:17, pp. 1763-1771).

But more recently, a study from Memorial Sloan-Kettering Cancer Center (MSKCC) in New York City concluded that while CT lung screening found more nodules and enhanced diagnosis and treatment, it "may not meaningfully reduce the risk of advanced lung cancer or death from lung cancer."

"Until more conclusive data are available, asymptomatic individuals should not be screened outside of clinical research studies that have a reasonable likelihood of further clarifying the potential benefits and risks," the report continued (Journal of the American Medical Association, March 7, 2007, Vol. 297:9, pp. 953-961).

One of the authors, MSKCC pulmonologist Dr. Peter Bach, said one risk to patients is the potential for unnecessary biopsies and surgeries. "If CT finds more nodules that pose little threat and triggers more treatment and more biopsies, then it can pose a greater threat."

"We had the same excellent survival after the removal of those cancers as other studies have reported," he continued. "There were lots of extra surgeries performed. Surgery has its own risk. Surgery mortality is about 4% or 5% in the U.S. Even so, there was no reduction in disease and no reduction in death. In other words, no benefit, just like chest x-ray."

No recommendations

The lack of conclusive benefits is the primary reason why no healthcare organization or association -- including the Society of Thoracic Radiologists, the American Cancer Society, or the American College of Chest Physicians -- recommends CT screening for lung cancer. And no third-party insurer reimburses for CT lung screening.

"There is the unresolved question about whether it is beneficial," said Dr. Charles White, director of thoracic radiology at Baltimore's University of Maryland Medical Center, who also does not recommend CT for lung screening. "I don't want to be in the position of recommending a screening test when there are two very strong opinions, each marshalling data that is unresolved. I'd rather wait to see what happens with the other data when it comes in."

In reviewing current studies, White said Henschke's research "has done a huge service of bringing the issue to the forefront of the scientific community and to the public at large."

He agreed that the key issue is whether CT can detect lung cancer at an early enough stage to make a difference in mortality, relating an analogy between lung cancer and prostate cancer. By the time men pass the age of 80, there is a great likelihood they will have prostate cancer, but chances are that they will not die of prostate cancer.

"The question becomes: Are these additional 1% or 2% of (lung) nodules ones that would have killed the patient, or would people live with them and have no problems?" White asked.

In addition, he cited Henschke's 1999 research, which detected noncalcified nodules in 23% of study participants through low-dose CT at baseline, compared with 7% by chest radiography. Malignant disease was detected in 2.7% by CT and 0.7% by chest radiography. Stage I malignant disease was found in 2.3% of the cases by CT and 0.4% by chest radiography (Lancet, July 10, 1999, Vol. 354, No. 9173, pp. 99-105).

Extrapolating the total to 100,000 patients, more than 2,500 lives could be saved by early screening, even with overdiagnosis of lung nodules.

Unlike prostate cancer, lung cancer is a disease patients overwhelmingly die of rather than with. Once lung cancer is diagnosed, which tends to be at later stages of disease, only about 13% of men and 16% of women are still alive five years later.

Managing nodules

"There is no doubt we can find small lung cancers with greater sensitivity than other screening modalities, such as chest x-rays," said Dr. Ella A. Kazerooni, director of the cardiothoracic radiology division at the University of Michigan (UM) Medical School in Ann Arbor. "The main problem remains the large number of false positives and how to manage those."

For every 100 patients with detected nodules, only one or two people may actually have lung cancer, Kazerooni said. The remainder of the patients still need their nodules monitored over time to ensure they are stable or, at some point, become large enough to require biopsy or surgical removal.

UM is part of the National Cancer Institute (NCI) Lung Image Database Consortium (LIDC) and one of the 46 centers participating in the National Lung Screening Trial (NLST). "Until we have data from the National Lung Screening Trial," Kazerooni added, "we won't have enough information to know if this is an efficacious screening."

The NLST is a cooperative initiative between the NCI and the American College of Radiology Imaging Network (ACRIN) to assess the efficacy of low-dose CT and chest x-ray in lung cancer screening.

NLST initiative

From 2002 to 2004, the NLST enrolled more than 54,000 men and women, ages 55 to 74, with a 30-pack-year history. (A pack year is calculated by one pack of cigarettes or another tobacco product per day, two packs a day for 15 years, etc.). Approximately half the participants were smokers at the time of enrollment, while the other half were former smokers who had quit within the previous 15 years.

The NLST is using multidetector CT scanners for screening, and performs phantom measurements to maintain low radiation dose and minimize -- if not eliminate -- the risk of carcinogenesis. This past February, the NLST completed its screenings for the study, with a participation rate of 90%. Follow-ups since have begun to determine how well the subjects are doing.

"We are concerned that there is a very high false-positive rate, people may have unnecessary surgical procedures, and that could lead to pulmonary complications and infections," said Dr. Christine Berg, NLST co-director and chief of NCI's early detection research group. "We are also assessing the impact on lung cancer mortality. Our next interim analysis will come up in (May)."

Berg is working with co-director Dr. Denise Aberle, who is also the director of thoracic imaging at the University of California, Los Angeles (UCLA).

Problematic diagnosis

While CT is beneficial at lung nodule detection, it is not a given that smokers and nonsmokers will naturally seek a cancer screening.

"Because the lungs do not have pain receptors per se, it requires that the cancer grow to a size large enough that it invades other structures for it to be symptomatic. That is why we detect it so late," said Dr. Geoffrey Rubin, chief of cardiovascular imaging at Stanford University in Stanford, CA. "Although we are picking up earlier-stage lung cancer, ultimately, from a public health perspective, we need to know if we are reducing mortality in these patients."

Rubin suggests that healthcare providers must be more effective in profiling patients and determining their risk level. "Maybe it results from genotyping; maybe it has to do with certain circulating biomarkers. By adding in extra filters, we might get to the point where we would enrich the population of positive results among those (people) being screened and not lose too many positives by not screening patients at lower risk," he said.

CT is unquestionably the modality of choice over x-ray in lung cancer screening. Rubin estimated that the odds of seeing a 1-cm lesion on a chest x-ray are 50-50, while a CT scan can routinely detect nodules 4 mm in diameter. "If the goal is detecting nodules, you want as effective a means as possible," he added. "I think the chest x-ray has already been shown to have failed as a screening test for lung cancer."

CT lung CAD

Computer-aided detection (CAD) technology may be of benefit for radiologists faced with the challenge of identifying small lung nodules in large volumetric datasets. CAD may be "a tremendous equalizer of performance across these readers," Run said. "Ultimately, we need to prove if CAD reduces cancer mortality, just like the questions related to screening in general."

2007 04 18 15 32 04 706
Ground-glass nodule, proven bronchoalveolar cell carcinoma; these are the hardest nodules for CAD to currently find. All images courtesy of Dr. Ella Kazerooni.

Kazerooni is active in the development of CAD software for CT lung screening at the University of Michigan and works with several vendors on their CAD CT lung products.

As CAD software helps reduce reader-to-reader variability in locating and measuring lung nodules and multislice CT provides more consistent volumetric data, CAD should enhance the detection and measurement of lung nodules, Kazerooni said. "From scan to scan, we are getting measurements that are correct, can accurately follow the growth rate of small nodules, and know which ones are the bad actors and which ones we don't have to worry about," she added.

2007 04 18 15 32 18 706
Above, a 5-mm noncalcified left upper lobe lung nodule, requires follow-up CT to look at growth pattern at six, 12 and 24 months. Below, same nodule above extracted in 3D using ALA software (GE Healthcare, Chalfont St. Giles, U.K.), giving a measurement of nodule volume as 67 mm3; changes in automated volume measurements may be more sensitive to growth than what a reader can identify visually on follow-up CT exams.
2007 04 18 15 32 30 706

Kazerooni also warned, however, that CAD software still is in what she described as a "preclinical" stage, and that CAD-like software from different sources must become more reliable.

On this issue, the Lung Image Database Consortium may prove helpful through its publicly available CT database. It will allow CAD developers to view lung nodule locations and sizes, and "test their CAD software against a proven gold standard and know if they are coming to the same conclusion," Kazerooni said. "Until CAD-like products can become consistent in the way they measure nodule volume, it will be hard to use them in general clinical practice."

By Wayne Forrest
AuntMinnie.com staff writer
April 19, 2007

Related Reading

CT screening for lung cancer: Implications on social responsibility, April 10, 2007

CT screening may not improve lung cancer survival, March 7, 2007

Italung-CT results show efficacy of lung cancer screening, March 9, 2007

Overdiagnosis common in lung cancer screening, February 7, 2007

CT lung cancer screening reduces mortality, October 26, 2006

Copyright © 2007 AuntMinnie.com

Page 1 of 653
Next Page