| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
Dr. Grannis is Head, Section of Thoracic Surgery, Department of General Oncologic Surgery, City of Hope National Medical Center.
Correspondence to: Frederic W. Grannis, Jr., MD, FCCP, Head, Section of Thoracic Surgery, Department of General Oncologic Surgery, City of Hope National Medical Center, 1500 E Duarte Rd, Duarte, CA 91010; e-mail: fgrannis{at}coh.org
In this issue of CHEST (see page 329), Jerome Reich, MD, from the Kaiser Foundation Hospitals Center for Health Research, adds his voice to the chorus of epidemiologists who oppose screening for lung cancer (LC) with CT scanning outside of a prospective randomized controlled trial (RCT). He marshals evidence, proposes models to elucidate the medical and economic perils of LC screening, and exhorts clinicians to rally round the "gold standard" of the RCT. The author provides a very complete review of previous screening trials for LC. He discusses the hypothetical reasons for failure of these trials to demonstrate a reduction in population mortality and concludes that overdiagnosis bias (ODB) has been convincingly demonstrated.
Ultimately, Reich’s theory, model, and conclusions are based entirely on indirect and circumstantial evidence. He offers little direct proof that ODB of LC exists. Furthermore, he ignores two clinical series that document very low survival in early stage LC patients who are not treated following diagnosis by screening.1 2 The information from the series by Sobue et al,1 as well as the data from the follow-up study by Flehinger et al2 of prior National Cancer Institute studies, indicates that most patients with small LCs who do not receive surgical resection or other effective treatment die within 5 years. Although this evidence is retrospective, it is very strong. Death is a clear end point.
Reich goes on to postulate that not only are 33% of LCs "pseudodisease" (ie, nonlethal if untreated), but that another 33% are "nonaggressive" in their behavior. This is completely at variance with the data from thousands of clinical series examining this disease. For example, national survival data indicate that only 15% of LC patients survive 5 years.3 Therefore, many patients must die of Reich’s "nonaggressive" LC. Any inferences drawn from such a misguided premise are, of necessity, invalid. Since epidemiologists insist that the RCT is a holy grail, then it is perfectly reasonable to insist that they drop discussion of putative "overdiagnosed" LC until they perform an RTC to demonstrate the existence and frequency of this ephemeral entity. I wish them good luck in finding a suitably compliant institutional review board and study subjects credulous enough to participate in the untreated control arm of their study.
Because resection of pseudodisease provides no benefit, Reich concludes that screening and treatment of early LC will actually result in a reduction in life expectancy. Why? Because the progressive decrease of FEV1 precipitated by surgery will worsen with age. He postulates that death occurs when FEV1 falls to 1 L. (This is inaccurate; many such patients will live for years with proper care.) He offers no RCT data to support this conclusion. One benefit of an early diagnosis of LC may be to preserve lung function. It is possible that future research in patients with very small, screen-detected LCs may show that some subsets may be effectively treated by limited resection or radiation therapy rather than by lobectomy.
It is difficult for me to understand how the RCT currently proposed by the National Cancer Institute can nominate chest roentgenogram as one arm, after reading Reich’s scholarly refutation of the value of this test. If one were intellectually and scientifically rigorous, one would have to insist on an RCT of CT scan against current medical practice (ie, diagnose and treat only after symptoms have developed). Could such a study be performed? I personally doubt that it could attract an adequate accrual of patients. Even with sufficient accrual, this RCT will require a minimum of 14 years to complete. During that time, millions will die and technology will change in ways that render current imaging techniques obsolete and the eventual study result meaningless.
Would this RCT be ethical? I do not think so. Randomization of a study subject or patient to a research study demands equipoise. The personal physician must conclude that he does not really know which of the two arms will offer his patient a better chance of survival. Would a reasonable physician make such a conclusion after looking at the data from the Early Lung Cancer Action Project (ELCAP) study, in which 80% of study subjects had LC detected in stage I?4 I seriously doubt it. As a former smoker who is at risk of LC death, my personal choice has been to become a study subject as well as a principal investigator in a participating center in the International-ELCAP (I-ELCAP) trial. Furthermore, if I did not have access to this trial, or was otherwise ineligible, I would have requested that the test be performed outside of a study by a skilled radiologic group. I could not, therefore, in good conscience, ask a study subject to risk a 50% chance of randomization to a study arm that was personally unacceptable. I suspect that many judicious clinicians will concur.
Reich also warns that not only will LC screening result in unnecessary deaths, it will also be ruinously expensive. I have addressed this issue in another editorial.5
Because there are so many billions of dollars involved in the implementation of LC screening, it is important to consider whether there are any potential conflicts of interest that might compromise unbiased debate. For example, radiologists and surgeons (like myself) might benefit financially from screening. Governments, medical insurance companies, and managed care organizations (like Kaiser) would be forced to absorb the costs. One major ancillary benefit to payers in an RCT is that any favorable result, and any consequent need to pay for LC screening, would be delayed for many years.
Another very interested third party is the tobacco industry. They recently obtained a favorable verdict in a class action suit in the state of West Virginia that sought to force tobacco companies to pay for CT scan and pulmonary function screening in 250,000 smokers and ex-smokers. They convinced jurors that LC screening was ineffective and dangerous.
While Reich is correct in insisting that there are important pitfalls in screening, it is important to put LC screening into a 21st century perspective. The lessons learned from prior screening trials in other organs have been carefully incorporated into the planning of the prospective single-arm I-ELCAP6 trial that is now in progress at 20 medical centers in the United States and a number of other nations. The I-ELCAP protocol carefully minimizes radiation dosage, maximizes quality control, fosters smoking cessation in study subjects, and incorporates a strict protocol emphasizing demonstrable growth in order to minimize the number of invasive tests and operations. The original ELCAP trial was very successful in all of these areas, and refinements in the protocol make I-ELCAP even better. The research plan of I-ELCAP will provide a definitive answer to the question of whether low-dose, noncontrast, spiral CT scanning will effect a stage shift toward early diagnosis. The overwhelming body of evidence available from hundreds of published surgical series suggests that surgical treatment in very small stage IA non-small cell lung cancer will result in high survival rates and should reduce LC population mortality. Paradoxically, the large scope of this multi-institutional trial also will offer a better opportunity to answer the question of ODB, since it is inevitable that there will be study subjects who are reasonably young and physiologically fit who will opt for no therapy. Careful follow-up in these individuals will provide a definitive answer as to the existence and frequency of pseudodisease LC and nonaggressive LC.
Finally, the performance of prospective single-arm trials does not mutually exclude the performance of large RCTs. There is definitely room for a reasonable difference of opinion over which type of study is superior. What is not acceptable is unwarranted overemphasis on highly theoretical risks while turning a blind eye on the very real carnage caused by LC. What is not acceptable is further pointless delay in implementing life-saving early detection strategies.
Footnotes
The author is a principal investigator in the City of Hope Lung Cancer Screening Project. This study is part of the I-ELCAP. Total grant support is $30,000 to provide a data coordinator to the protocol.
References
This article has been cited by other articles:
|
|
 |
|
  J. A Dempsey, C. A Smith, T. Przybylowski, B. Chenuel, A. Xie, H. Nakayama, and J. B Skatrud The ventilatory responsiveness to CO2 below eupnoea as a determinant of ventilatory stability in sleep J. Physiol., October 1, 2004; 560(1): 1 - 11. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Esteva Surgery in Lung Cancer Patients Chest, August 1, 2004; 126(2): 656 - 656. [Full Text] [PDF]
|
|
|
|
|
|
F. W. Grannis Jr Lung Cancer Screening: Response to Jerome Reich, MD Chest, January 1, 2004; 125(1): 350 - 351. [Full Text] [PDF]
|
|
|
|
|
|
J. Reich Re: Contumacy vs Mendacity Revisited; Response to Dr. Grannis' Letter Chest, January 1, 2004; 125(1): 351 - 352. [Full Text] [PDF]
|
|
|
|
 |
|
 American Thoracic Society/European Respiratory Society Statement: Standards for the Diagnosis and Management of Individuals with Alpha-1 Antitrypsin Deficiency Am. J. Respir. Crit. Care Med., October 1, 2003; 168(7): 818 - 900. [Full Text] [PDF]
|
|
|
|
|
|
J. M. Reich Lung Cancer Screening: Contumacy vs Mendacity Chest, March 1, 2003; 123(3): 963 - 964. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
