| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
a Lung Cancer and Thoracic Oncology Program, City of Hope National Medical Center, 1500 E. Duarte Rd, Warsaw MOB, Duarte, CA 91010-3000
b Medical Data Research Center, Providence Health System, 44785 NW Elk Mountain Rd, Banks, OR 97106
(Email: kkernstine{at}coh.org).
In the 1990s, fluorodeoxyglucose positron emission tomography (FDG-PET) was introduced as a part of the evaluation of suspicious solitary pulmonary nodules, and visual assessment and the maximum standard uptake value (SUVmax) pixel in the region of interest have been used to estimate the likelihood of malignancy [1, 2]. Most studies have concentrated on larger lesions of 3 cm or more [3]. FDG-PET is unlikely to be as accurate for the smaller lesions. So, the question asked by Ohba and colleagues [4]—can the accuracy of FDG-PET be improved in evaluating malignancy in lesions less than 3 cm?—is certainly relevant.
Others have attempted to improve the accuracy of PET by manipulating the emission results [2, 5]. Ohba and colleagues, like Nomori and colleagues [6], derived their calculations without demonstrating the physiologic reason why the calculations might be better than visual assessment or SUVmax to assess the solitary pulmonary nodules. The PET data manipulations may not be generalizable to other patient groups or FDG-PET systems: the machines, FDG injected, the FDG-PET protocols, analytic software, and the readers differ across institutions. There are no consistencies.
There appears to be several areas of concern in which bias may have been introduced:
These are a few of the reasons that we are concerned with the results.
Some statistical considerations should be noted as well. First, without a clear reason for the manipulation of the PET numbers, the PET data could be recalculated innumerable ways—and the estimation of malignancy improved—in a retrospective fashion. Because there are numerous hypotheses and several ways of calculating a "cutoff," the number of patients necessary will need to be significantly larger, several hundred [7]. Second, the authors show their receiver operating characteristic (ROC) curves but do not provide us with the raw data that produced them or the area under the curves for our comparisons. Finally, they derive "optimal cutoff values" for their ROC curves, but no method is presented and "optimal" is not defined. It is better to use some clinical criteria to select the optimal specificity and sensitivity. A recent recommendation is that at least 100 true-positives in 100 true-negatives are necessary to obtain the optimal number [8], far fewer than what are present than in the Ohba article.
In conclusion, although it is an interesting approach to use the FDG-PET data in a different fashion than visual assessment and SUVmax, the concerns about the patient population and the method of analysis make the results suspect. Perhaps a larger patient population, possibly from more than one institution, with very clear definitions and a prospective study protocol, will help us to have greater confidence in these results.
 |
References |
|---|
 Top References |
|---|
Related Article
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| ANN THORAC SURG | ASIAN CARDIOVASC THORAC ANN | EUR J CARDIOTHORAC SURG |
| J THORAC CARDIOVASC SURG | ICVTS | ALL CTSNet JOURNALS |
