Researchers defend NLST protocol against its critics

January 1, 2008

The first interim results from the first study of dasatinib (Sprycel) as induction therapy of Ph+ acute lymphoblastic leukemia (ALL) show complete and early hematologic responses with good overall compliance

High-risk individuals who have lung cancer detected by low-dose spiral CT screening have longer survival than people with clinically detected lung cancer, and two large, randomized studies are seeking to determine whether such screening also reduces lung cancer mortality. But will one trial provide more conclusive evidence than the other? Is the US study, the National Lung Screening Trial (NLST), inferior to the European study known as NELSON, as some critics have claimed? Not in the view of lead researchers involved in both studies. They regard the trials as complementary, not competitive.

"Although it is interesting to compare the two trials, you should realize that a comparison in terms of superiority/inferiority is probably inappropriate," NELSON principal investigator Rob J. van Klaveren, MD, PhD, of Erasmus Medical Center, Rotterdam, told ONI. "The two studies have different designs and different approaches. They are, in a sense, complementary."

In a separate interview, NLST co-director Denise R. Aberle, MD, of UCLA, called the NELSON trial extremely important. "When our data are mature," she said, "we will be able to consider methods of doing statistical meta-analysis. Our patient cohorts are close enough that we will gain information and, in effect, validation from the NELSON trial because it was done by an independent group."

NLST critics consist largely of advocates for low-dose CT screening who are convinced that earlier I-ELCAP reports showing increased survival (N Engl J Med 355:1763-1771, 2006) are adequate to support its use in examining high-risk individuals for early lung cancer.

"There are groups that go to state legislatures to try to get money from the Tobacco Settlement funds apportioned to researchers who are pro screening and to the US Congress to get legislation introduced to mandate Medicare coverage for screening," said Peter Bach, MD, of Memorial Sloan-Kettering Cancer Center, who is not involved in NLST.

These same individuals and groups appear to be trying to find flaws in the NLST, he said, perhaps because they fear the results.

"The research community regards a randomized trial of screening to be the gold standard, and there is a legitimate chance that NLST may show that CT is not beneficial," Dr. Bach said.

The study protocols

NLST is supported by the National Cancer Institute and conducted by the Institute in collaboration with the American College of Radiology Imaging Network (ACRIN). The two groups use the identical protocol for their primary and secondary endpoints. In addition, ACRIN sites are carrying out additional studies under their NCI contract using the same patient population.

The NLST researchers enrolled 53,464 individuals at high risk of lung cancer (59% male) from 2002 into 2004 at more than 30 sites in 28 states. Participants were randomized 1:1 to receive spiral CT or chest x-ray. Patient follow-up will continue at least through 2009, and could last 9 years. The trial is powered to identify a 20% mortality difference between the two arms by the end of 2008.

NELSON researchers, whose acronym comes from the trial's name in Dutch, enrolled about 20,000 high-risk participants (85% males) beginning in 2003 at five screening sites in The Netherlands, Belgium, and Denmark. Unlike the NLST, its control group gets only usual care. NELSON is scheduled to conclude at the end of 2015, and it is powered to detect a 25% mortality difference.

An obvious difference between the two trials, and one source of criticism of the NLST study, is the US study's use of chest x-rays in the control group.

"Some people have criticized the NLST because of the chest x-ray," said NLST investigator William C. Black, MD, director of chest radiology at Dartmouth-Hitchcock Medical Center's Norris Cotton Cancer Center. "I don't think it will make a difference."

Simply put, the argument against chest x-rays is that study members in the control group who are referred for further evaluation of abnormal lung findings will likely receive low-dose CT imaging, per study guidelines. If too many in the control arm receive CT scans, this will contaminate the findings and make it difficult to detect the true difference in mortality between the NLST's intervention and control arms.

Not so, Dr. Aberle argued. "A follow-up of a possible lung cancer abnormality on chest x-ray is not equivalent to contamination," she said. "If we are trying to determine the incremental benefit of CT relative to chest x-ray, it doesn't make any difference if an abnormal chest x-ray eventuates in a diagnostic CT scan because the lung cancer found by a follow- up CT is based on that abnormal chest x-ray. So the claim that it is a contamination is incorrect."

The Lung Cancer Alliance, a strong proponent of CT screening, has also implied that the scanning technology used in NLST is outdated.

"It is uncontestable that the NLST has had superlative image quality using the most advanced available technology," Dr. Aberle countered.

She noted that the study "mandated only multidetector scanners that had a minimum of four detector rows. Each scanner that was incorporated into the trial underwent an independent evaluation by a CT physics group to determine the most appropriate perimeter for obtaining or completing that examination, such that there would be a consistent level of image quality, spatial quality, and resolution quality across the trial."

Too small?

Is the size of the NLST study too small to achieve its primary goal of detecting a mortality difference between its two arms, as a few critics have contended? Even the question is derided by people inside and outside the trial.

"If it's not large enough to detect a difference, the only reason could be that the difference is so microscopic that it couldn't possibly be clinically important," Dr. Bach said.