NLST article not balanced, critic of screening trial asserts

February 1, 2008

"Researchers defend NLST protocol against its critics" (ONI, January 2008, page 14) quoted three individuals in the United States who strongly support the NLST [National Lung Screening Trial] and no one on the other side of the issue.

 

The story "Researchers defend NLST protocol against its critics" (ONI, January 2008, page 14) quoted three individuals in the United States who strongly support the NLST [National Lung Screening Trial] and no one on the other side of the issue.

The article should have noted that two of the NLST investigators quoted (Drs. William C. Black and Denise R. Aberle) have taken money from tobacco companies for expert witness testimony or reports in medical monitoring lawsuits designed to compel the tobacco industry to pay for lung cancer screening of their customers/victims.

Peter Bach, MD, who was also quoted, should have been identified as a senior consultant to the director of the Centers for Medicare & Medicaid Services (CMS), the government entity that would have to pick up the check for lung cancer screening of high-risk individuals if it became standard practice.

 

Problems with the control arm

A more balanced news story would have pointed out that the randomized controlled study is only "golden" when it is properly designed. The NLST is not properly designed.

The experimental intervention (low-dose spiral CT in this case) should be compared with the current standard (zero screening) as in the NELSON trial, rather than an intervention that is not currently utilized, ie, chest x-ray (CXR).

The National Cancer Institute is crystal clear in its repeated recommendation that chest roentgenograms are considered ineffective and should not be used for population lung cancer screening. The NLST design improperly violates this principle and utilizes CXR as a screening arm.

This is bad science! And contrary to Dr. Black's assertion, it does make a major difference in the statistical power of the NLST trial. In fact, it may guarantee a false-negative result.

In study subjects who get no screening, as in the NELSON trial, the primary study measure, lung cancer specific mortality, is expected to be about 85% at 5 years.

In the Mayo Lung Trial, the survival of patients with lung cancer detected by CXR was 34% at 5 years. In more modern studies from Japan, lung cancer survival at 5 years is closer to 50% in CXR-screened patients and approximately 80% in CT-screened patients.

This means there will probably be a much smaller survival difference between CT and CXR (ie, 30%) than between CT and no screening (65%). This makes it much more difficult to find a statistically significant difference in lung cancer mortality between CT- and CXR-screened arms than between CT-screened and unscreened arms.

 

Follow-up too short?

This is only the first problem with NLST. In ONI's article, the NLST researchers say they expect to publish results in 2009. This means there will be only a maximum 5-year follow-up of screened individuals in the study.

But the minutes of the Data Safety Monitoring Board of the NLST clearly show that validation of cause of death is only obtained in half of the study deaths, and, in addition, the delay in determining the cause of death is 23 months! This means that the actual follow-up in the NLST will only be about 3 years.

Thus, the NLST is not just badly designed, the conduct of the study is also deficient.

The Japanese ALCA [Anti-Lung Cancer Association] study adds a lead-in period of 2 years to represent the screening phase of the study druing which mortality is very low. If one takes the survival curves from the ALCA study and makes a mirror image to reflect probable mortality curves, it becomes immediately evident that the difference in mortality between CXR and CT widens over time and would only exceed 20% beyond 5 years.

In this context, one should expect to see just about a 20% difference in mortality between CT screening and CXR screening at the end of the NLST.

The inescapable conclusion of these considerations is that there is a strong possibility that the NLST will be a false-negative study, despite a major difference in long-term mortality, if it is published too early, ie, in 2009.

 

Who benefits from a negative result?

The biggest beneficiary of such a negative result will be the tobacco companies, who can ask their expert witnesses to cite this study in courtrooms during medical monitoring lawsuits in New York, Massachusetts, and perhaps elsewhere. (I have provided testimony against the tobacco companies in these actions.) If Big Tobacco wins these cases, as they did with the help of Dr. Aberle in Louisiana, it will save them billions of dollars.

Also benefiting will be CMS (see above) and health care insurers who will not have to pay for screening.

The clear losers will be innumerable lung cancer patients who will continue to suffer and die unnecessarily, despite the fact that we now have a screening method and protocol that has been proven in a large prospective multi-institution international research study (I-ELCAP) to allow the lives of 80% of patients with CT-detected lung cancer to be salvaged by surgery without the need for adjuvant radiation or chemotherapy.

This comment addresses only a few of the many weaknesses of the NLST trial. There are many other problems with NLST that experts would be able to explain to your reporters, if they would only ask.

 

- Frederic W. Grannis, Jr., MD
Department of Thoracic Surgery
City of Hope National
Medical Center
Duarte, California