| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Department of Radiology, Weill Cornell Medical Center 525 East 68th Street, New York, NY 10021. e-mail: dyankele@med.cornell.edu
Editor:
In the March 2003 issue of Radiology, Dr Swensen and colleagues (1) presented results from their study on computed tomographic (CT) screening for lung cancer. In the discussion section of the article, they noted, "our results both raise hope that screening CT could be an effective tool to decrease mortality from lung cancer and raise concern that false-positive results and potential for overdiagnosis at screening CT could result in more harm than good" (1).
They go on to suggest how we may learn the truth about the hope and the concerns. To Dr Swensen and colleagues, a randomized controlled trial such as the National Lung Screening Trial will help provide answers: "A randomized controlled trial that shows disease-specific mortality benefit may be the best way to settle the controversy" (1).
In their recent study, Dr Swensen and colleagues have already learned a good deal about the small nodules that are found with initial CT screening for lung cancer. On the basis of their results, modifications to the diagnostic work-up of detected nodules have already been made in the protocol of the National Lung Screening Trial. Similarly, on the basis of our results in the Early Lung Cancer Action Project, or ELCAP, we have modified our approach to small nodules at baseline screening, and we have thus been able to reduce the rate of false-positive findings from the initial test at baseline screening by 60%. My point is that we only learn about these small nodules through experience with screening. We need to continue to learn how to manage these small nodules diagnostically with the goal of minimizing work-up without undue failures in the detection of cancer, and we need to learn about the prognostic significance of the "cancers" diagnosed in subsolid nodules in particular.
As for the overdiagnosis issue, Dr Swensen and colleagues describe the results from the Mayo Lung Project as demonstrating the presence of overdiagnosis even in the context of chest radiographic screening. I have reviewed the original data from the Mayo Lung Project, and if one were to take seriously the concern for overdiagnosis in chest radiographic screening, as Dr Swensen and colleagues do, then more than half of stage I cancers detected at screening would have been considered overdiagnosed. These cancers were typically about 2 cm in diameter at diagnosis, however; they were not seen at baseline screening but only at repeat screening; and they grew with rates typical of symptomatic malignancies (2). Thus, they were not likely candidates for overdiagnosis.
More important, however, is our need to learn which subtype(s) of "cancer" overdiagnosis occurs at screening and with what frequency. Again, Dr Swensen and colleagues have already learned from their recently reported study that the likely candidate for overdiagnosis in CT screening is "cancer" in a nonsolid nodule. The National Lung Screening Trial does not provide this type of information.
I am deeply concerned about the commitment that radiologists have made to randomized controlled trials as a means of evaluating screening for cancer (3). Frequently, issues such as those raised by Dr Swensen and colleagues are used to justify randomized controlled trials, even though the trials shed no light on those issues at all. The academic radiology community should seriously review this issue in an open discussion among experts so that we can avoid the type of confusion that has ensued with respect to mammography (4). I suggest that an academic journal such as Radiology take a leading role in organizing this discussion. After all, it was the radiology community itself that introduced the idea of randomized controlled trials in diagnostic research (with diagnostic testing viewed as an intervention), and it is thus the radiology community’s responsibility to recognize and correct its mistake.
REFERENCES
Department of Radiology, Mayo Clinic, 200 First Street SW, Rochester, MN 55905
We appreciate Dr Yankelevitz’s thoughtful letter to the editor. Our response follows regarding his queries:
In our ongoing National Cancer Institute study, we continue to describe and recommend management for all lung nodules, regardless of size. We currently recommend annual follow-up for the smallest nodules (ie, smaller than 4 mm in diameter).
The large number of false-positive findings associated with CT screening for lung cancer will be a challenge to overcome, although it may not be insurmountable. To date, more than 72% of our participants have a radiologically indeterminate uncalcified lung nodule. More than 98% of these radiologically indeterminate uncalcified lung nodules are benign and are therefore false-positive findings. Dr Yankelevitz states that they have been able to "reduce the rate of false-positive findings at baseline by 60%." Modification of a definition of or the approach to small nodules at baseline screening does not reduce the rate of false-positive findings; it simply reduces the follow-up required for them or ignores them completely. Fewer follow-up examinations lead to reduced expense, which is desirable, but these ignored nodules will include some cancers. In our group, there were four cancers that measured 3 mm in diameter at detection. Actual reduction of the false-positive rate would require changing the image acquisition so that the small nodules are not seen.
We discuss the study bias of overdiagnosis thoroughly in our article. This is a very real concern of many researchers in this area. We outline evidence from Japan and the United States that strongly implicates a portion of lung cancers as pseudodisease, which results from overdiagnosis bias. This is not an uncommon bias, and evidence suggests that it plays a role in prostate, renal cell, and breast carcinomas. It becomes even more of an issue in patients with a heavy smoking history, given the comorbidities of chronic obstructive lung disease, heart disease, and propensity for stroke, because they are even more likely to die of real disease a very slow growing lung cancer (pseudodisease).
The radiology community’s commitment is required to solve puzzles, such as determining appropriate nodule evaluation and what to do with small ground-glass cancers, but these will not provide effectiveness for CT screening. Some of our best and brightest scientists and physicians at the National Institutes of Health have determined, along with experts throughout the world, that the best way to address the question of whether CT screening reduces mortality from lung cancer is through a randomized controlled trial. The National Cancer Institute has appropriated approximately $200 million to address this issue, with a well-thought-out randomized controlled trial called the National Lung Screening Trial. We encourage each of you to help in recruitment to this study of 50,000 high-risk Americans to answer this important public health question for the United States and the world as soon as possible.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
