| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Department of Radiology, New York Presbyterian Hospital, Building Starr-JO30, 525 E 70th St, New York, NY 10021
e-mail: dyankele{at}med.cornell.edu
The article by Dr Lindell and colleagues (1), published in the February 2007 issue of Radiology, concludes that "overdiagnosis, especially in women, may be a substantial concern in lung cancer screening."
The basis for this conclusion was that 13 (27%) of the cancers had doubling times of longer than 400 days. The abstract of the article reports only the mean tumor volume doubling time, which was 518 days. Further review of the results shows that the median doubling time was 166 days, and the range was 10–5810 days. Overall 73% had doubling times shorter than 400 days, and among men there were only two cases (8%) of doubling times longer than 400 days.
The measurement techniques (bidimensional measurements) used in the study were imprecise. In addition, many of the slow-growing tumors were not of solid consistency, and the techniques for measuring volume change in nonsolid and part-solid nodules are not well defined. As noted, many of these tumors changed in terms of their consistency, with some becoming completely solid, and this can be viewed as representing a change in volume.
In regard to the results of this study, it would have been more correct to report as the primary result the median doubling time of 166 days, since the average is so skewed by the outliers. Overall, the article reports that, at a minimum, 73% of the screening detected cancers are in fact genuine, especially among the men. Surely finding cancers with these rapid growth rates when they are small and asymptomatic is far better than waiting until they produce symptoms.
These issues aside, the concept of quantifying overdiagnosis by use of growth rate is very useful. It demonstrates that overdiagnosis does not need to be thought of as some sort of "screening bias" only to be studied in the context of screening randomized controlled trials (RCTs). In the future other techniques, perhaps biologic ones, will become available. The issue therefore becomes not one of overdiagnosis but rather one of overtreatment, with the idea to learn which subtypes of tumors are the slowly progressive ones so as to factor that information into the treatment plan. Notably, this type of information, regarding which subtypes to focus on, cannot be learned as a consequence of screening RCTs.
An additional question that also arises from these results is why was surgery performed in patients with doubling times as long as 5810 days? The Mayo group has previously reported surgery occurring in the screening population yielding benign results in over 18% of the cases (2). This raises concern regarding the work-up algorithm used in that study.
The author is a consultant for PneumRx (Mountain View, Calif) and receives royalties from Cornell University from a licensing arrangement for patented technology with General Electric.
 |
References |
|---|
 TOP References References |
|---|
,
James R. Jett, MD ,
Thomas E. Hartman, MD *, and
Stephen J. Swensen, MD *
* Department of Radiology, Mayo Clinic, Charlton 2-290, 200 1st St SW, Rochester, MN 55905
e-mail: Lindell.Rebecca{at}mayo.edu
Department of Pulmonary and Critical Care Medicine, Mayo Clinic, Charlton 2-290, 200 1st St SW, Rochester, MN 55905
We concur that bidimensional measurement may be less precise than a three-dimensional volumetric measurement; however, at this time bidimensional measurement remains the primary means of nodule measurement in practice. We also agree that ground-glass opacities may increase in density without increasing in size, and this makes calculation of traditional doubling times problematic.
We disagree that it is more correct to report as a primary result a median doubling time of 166 days. Both this and the mean doubling time of 518 days are correct and were reported in the article. The focus of the article was on the issue of overdiagnosis and this is why the mean doubling time was emphasized.
Our study (1) was not designed to address the issue of whether computed tomographic (CT) detection was better than symptom detection. We are pleased that Dr Yankelevitz recognizes that overdiagnosis is a reality with CT screening. Given that these slowly growing cancers would likely cause neither symptoms nor death and that they would not be detected in the absence of CT screening, they are certainly a manifestation of screening bias. Although overdiagnosis leads to over treatment, we do not believe that turning the discussion to "overtreatment" adds clarity.
Our data suggest that at least 25% of cancers detected with CT have a growth rate that may not be lethal if left untreated. Considering the competing causes of death among long-term smokers, the percentage of lung cancers that would not be the cause of death if untreated may be much higher. To the contrary, these data lend further, compelling support for the need for an RCT to establish efficacy for CT screening. A single-arm observational screening program cannot measure presence or absence of mortality reduction or the effect of overdiagnosis bias on apparent improvement in survival.
We agree that surgical intervention for benign diagnoses is best kept to a minimum. Patients make an informed decision regarding slowly growing nodules, and many patients opt for removal. We have been forthright in publishing the number of benign diagnoses attained at surgery, and this rate is comparable with those of other screening series and is much lower than those of many historical series (2,3). These benign nodule resections represented 0.3% of all nodules detected with screening.
In the end, overdiagnosis with or without treatment has an effect on the patient and needs to be taken into account in the evaluation of the effectiveness of CT screening studies.
|
References |
|---|
|
TOPReferencesReferences |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
