| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
and
Ulrich Buell, MD*
Department of Nuclear Medicine* and Institute of Medical Statistics, University Hospital Aachen, Pauwelsstrasse 30, 52074 Aachen, Germany. e-mail: preinartz@compuserve.com
Editor:
In their article in the December 2003 issue of Radiology, Dr Coche and colleagues present a comparison between the accuracies of pulmonary computed tomographic (CT) angiography and ventilation-perfusion lung scintigraphy in the diagnosis of acute pulmonary embolism (1). Such a study is of high interest, since increasing competition between the two imaging modalities can be observed.
Unfortunately, the authors missed the opportunity for a well-balanced and impartial evaluation. In their study, an advanced imaging modality—thin-collimation multi–detector row spiral CT—was compared with a planar modality performed with a technique similar to that of the Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED) study, which was designed in 1983. None of the recent scintigraphic advances in acquisition technique, image reconstruction, or image analysis were taken into account. The minimum requirements for a balanced study design would have been tomographic acquisitions of ventilation and perfusion by means of single photon emission CT (SPECT).
Nowadays, SPECT is a well established and ubiquitously available imaging method that is widely used in modern nuclear medicine diagnostics. Even the acquisition time is no longer a methodic weak point of SPECT: Provided that a double- or triple-head camera is used, the acquisition time of a SPECT scan is comparable to that of planar imaging (2,3). Other technical advances that should also have been taken into consideration are iterative image reconstruction algorithms and modern aerosols (2,4).
In addition, the statistical analysis seems to be debatable. With regard to ventilation-perfusion scintigraphy, the cases of intermediate class were considered positive for embolism when the specificity was calculated, while the very same cases were considered negative for embolism when the sensitivity was calculated. By doing so, the statistically worst achievable results were published for scintigraphy. A calculation of the sensitivity and specificity in which the cases of intermediate class are consistently considered either positive or negative renders values of 96% and 88% or 86% and 94%, respectively. These values are distinctly better than those of 86% and 88%, which are provided in the article. The fact that these data were mixed up in the abstract leads to further confusion.
In conclusion, we would like to express our opinion that a balanced and impartial assessment of the accuracy of CT angiography and ventilation-perfusion lung scintigraphy in the diagnosis of pulmonary embolism is overdue. However, such a study should lead to an unbiased and impartial comparison between the two imaging modalities, each performed with state-of-the-art technique.
REFERENCES
,
André Keyeux, MD, PhD,,
Franck Verschuren, MD,
and
Francis Zech, MD
Departments of Medical Imaging,* Nuclear Medicine, and Emergency Medicine, Cliniques Universitaires St-Luc, Avenue Hippocrate 10, Brussels B-1200, Belgium. e-mail: coche@rdgn.ucl.ac.be
We appreciate the interest of Dr Reinartz and colleagues concerning our article (1). We would like to respond to their comments as follows.
In response to the first point, concerning the technical characteristics of planar ventilation-perfusion scintigraphy, we maintain that the planar modality used in our study cannot be described as similar to that of the PIOPED study. We should perhaps emphasize once again that the reference to PIOPED is obsolete because ventilation in the PIOPED study is performed with xenon gas, and regional washout measurements are obtained with use of a single dorsal position. In our study, krypton gas was used to allow (a) the recording of as many ventilation images as perfusion ones and (b) a valuable comparison between perfusion and ventilation. Furthermore, the criteria used for the evaluation of the perfusion-ventilation test were not those of PIOPED, as clearly mentioned in the materials and methods section.
With regard to up-to-date scintigraphic techniques, we agree with Dr Reinartz and colleagues that SPECT could be an interesting modality with which to compare results. Moreover, some colleagues of our institution obtained encouraging results with the use of perfusion SPECT alone (2). Therefore, we have for the past 2 years decided to systematically perform perfusion SPECT in our population. At the present time, we are analyzing the performance of perfusion SPECT for future publication. We preferred to first present the results obtained with the planar technique because, to our knowledge, there are no publications to date that provide clear and validated criteria for the interpretation of SPECT findings and because we know that if we perform simultaneous ventilation and perfusion SPECT with krypton 81m (81mKr) as the ventilation agent, we will need to correct for the cross talk of the 81mKr in the technetium 99m window, cross talk that can vary according to the angular view and craniocaudal distribution (3). We also agree with Dr Reinartz and colleagues concerning their remark about the use of technegas. Unfortunately, we do not have a detailed experience with technegas, but we know that the use of this agent can be problematic in patients with severe illness and may increase the number of nondiagnostic scans (4).
With regard to the statistical analysis, we thank Dr Reinartz and colleagues for pointing out the error in our abstract (1). An erratum appears in this issue.
The sensitivity of thin-collimation multi–detector row CT and ventilation-perfusion scintigraphy for the detection of pulmonary embolism was 96% (27 of 28 patients; 95% CI: 82%, 99%) and 86% (24 of 28 patients; 95% CI: 67%, 96%), respectively. The specificity of CT and scintigraphy was 98% (65 of 66 patients; 95% CI: 92%, 99%) and 88% (58 of 66 patients; 95% CI: 77%, 94%), respectively.
We applied the same analysis for both diagnostic tests, taking into consideration inconclusive results for statistical purposes for both modalities. Sensitivity was defined as the probability for the test results to be positive in the presence of disease, and specificity was defined as the probability for the test results to be negative in the absence of disease. In the present study, positive CT and high-probability ventilation-perfusion scintigraphic findings were considered to represent a positive test result, while normal or near-normal scintigraphic and negative CT findings were considered to represent a negative test result. Indeterminate CT (n = 1) and intermediate-probability scintigraphic (n = 7) findings were considered essentially worthless and necessitated further diagnostic work-up in the present study. Pulmonary angiography was then attempted and was used as the method of reference to confirm or exclude pulmonary embolism.
For the measurement of sensitivity, we considered that in patients with proved pulmonary embolism (n = 28), intermediate-probability scintigraphic examinations (n = 3) could not be regarded as having positive results because they failed to demonstrate the disease with certainty. For measurement of specificity, we considered that in patients without proved pulmonary embolism (n = 66), intermediate-probability scintigraphic (n = 4) and indeterminate CT (n = 1) findings could not be regarded as negative findings because these results warranted further diagnostic work-up to rule out the disease definitively.
As suggested by Dr Reinartz and colleagues, we can perform the exercise of replacing intermediate scintigraphic results with either positive or negative results in our analysis, and when doing so, we obtain a sensitivity of 96% (positive results) and 86% (negative results) and a specificity of 88% (positive results) and 94% (negative results) for scintigraphy. Some of these results are indeed better than those presented in our article, but Dr Reinartz and colleagues postulate that intermediate-probability scintigraphic examinations have either positive or negative findings. In clinical practice, we cannot accept this postulate because intermediate scintigraphic results are nondiagnostic and should not be an end point in the evaluation of pulmonary embolism. Clinicians will not institute anticoagulation or other therapies on the basis of such examination results.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
