| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Editorial |
1 From the Department of Radiology, University Hospital, Rm 0279, 550 N University Blvd, Indianapolis, IN 46202-5253. Received December 15, 1999; accepted December 20. Address correspondence to the author (e-mail: matjohns@iupui.edu).
Index terms: Computed tomography (CT), angiography • Computed tomography, comparative studies, 99.124, 95.12916 • Computed tomography (CT), helical, 957.12912, 957.12915, 957.12916 • Editorials • Hypertension, portal, 957.711 • Portography, 95.124 • Shunts, portosystemic, 95.453, 957.1268 • Stents and prostheses, 957.1268
Transjugular intrahepatic portosystemic shunt (TIPS) creation has proved to be effective in the treatment of variceal bleeding and refractory ascites due to portal hypertension. Shunt stenosis is a frequent complication that, if not treated, can lead to shunt thrombosis and recurrent bleeding or ascites. Early detection and treatment of shunt dysfunction decreases the incidence of these complications. Portography with portal manometry is the recognized standard for the diagnosis of lesions causing shunt dysfunction but is, of course, invasive (1).
Ultrasonography (US) has been demonstrated to be an excellent method of screening for shunt stenosis due to its relatively high diagnostic accuracy, low cost, lack of ionizing radiation, and noninvasive nature. Chopra et al (2), in this issue of Radiology, describe their evaluation of helical computed tomographic (CT) angiography as an alternative modality for the evaluation of shunt patency in patients with TIPS.
Dr Chopra and his colleagues are to be congratulated for the quality of the study that they performed. They evaluated not only the ability of helical CT angiography to demonstrate morphologic shunt abnormalities but also its ability to allow judgment of the hemodynamic significance of those abnormalities. The latter is extremely important in the evaluation of a potential screening modality: The ability of a modality to depict a morphologic abnormality is not as important as its ability to demonstrate whether that abnormality compromises shunt function. Further, the authors' analysis of the relative contributions of several forms of data presentation to that demonstration is commendable.
The authors evaluated the diagnoses submitted by the reviewers for various combinations of the transverse source, multiplanar reformation (MPR), and opacity-assigned images and for each form of image alone. They achieved 97% sensitivity, 89% specificity, and 94% accuracy in the demonstration of morphologic abnormalities and respective values of 92%, 77%, and 84% in the demonstration of hemodynamically significant abnormalities when diagnoses were based on a review of all three modalities (transverse source, MPR, and opacity-assigned).
Those high sensitivities are laudable and are of more clinical import than are the relatively lower specificities: While a false-positive finding would cause the patient to undergo direct portography and manometry (ie, standard evaluation), a false-negative finding might allow a significant stenosis to go untreated, with resultant shunt thrombosis and variceal bleeding or recurrent ascites. Unfortunately, the high diagnostic accuracy achieved by means of evaluation of the transverse source, MPR, and opacity-assigned combination was accompanied by relatively poor interobserver agreement (
= 0.62 for the demonstration of all abnormalities, = 0.54 for the demonstration of significant abnormalities).
Fortunately, inclusion of the opacity-assigned images in the interpretation was not necessary. Diagnoses achieved with interpretation of images in the transverse source and MPR combination were nearly as accurate as those obtained with the transverse source, MPR, and opacity-assigned combination: The accuracy of the transverse source and MPR combination in the definition of morphologic abnormalities was only slightly less (92%), with no decrease in sensitivity (the more important parameter in this setting) and with only one additional false-positive finding. That single additional false-positive finding also decreased the accuracy of the interpretation of images in the transverse source and MPR combination in the demonstration of significant abnormalities, but only slightly, from 84% to 82%.
Of greater importance than the apparent decreases in accuracy of this combination is the associated superior interobserver concordance ( = 0.75 for the demonstration of all abnormalities, = 0.84 for the demonstration of significant abnormalities). Thus, removal of the interpretation of images achieved with proprietary software (opacity-assigned images) had no effect on the sensitivity of the putative screening examination and increased agreement among reviewers. This benefit, in addition to the greater familiarity of practicing radiologists with transverse source and MPR images than with opacity-assigned images supports the use of the transverse source and MPR combination in image interpretation.
The authors have clearly demonstrated that helical CT angiography is capable of demonstrating TIPS abnormalities. A question is raised by that demonstration: Should helical CT angiography be used in TIPS surveillance, or should the modality currently used in most centers, US, remain the modality of choice? The answer to that question requires evaluation of the strengths and weaknesses of both modalities.
Chopra et al refer to the study by Kanterman et al (3), who demonstrated that the sensitivity and specificity of US in the detection of TIPS abnormalities is 92% and 72%, respectively; these values are similar to those of helical CT angiography (transverse source and MPR combination) for the demonstration of significant abnormalities (92% and 73%).
Other well-designed studies (4,5) have demonstrated similar diagnostic accuracy for US in this setting. As noted by Haskal et al (5), the sensitivity and specificity of US are dependent on the location at which blood velocity is measured (eg, midshunt versus portal vein) and on the arbitrary velocity below which an abnormality is deemed present. Hence, US is operator-dependent. By eliminating this dependence, helical CT angiography could conceivably prove to be a less subjective modality than US. However, despite that potential lesser subjectivity, no improvement in the diagnostic accuracy of one modality over another can be stated. Indeed, the diagnostic accuracy of both modalities in this setting appear to be equivalent.
In support of the use of helical CT angiography, Chopra et al note that US can be "time-consuming and difficult to perform" and provide examples of "the poor window provided by the cirrhotic liver, deep location of the stent, difficulty in obtaining an adequate angle of insonation, and presence of ascites make Doppler US of TIPS difficult to perform, even by experienced sonographers." Of course, they are correct.
There are countless scenarios in which US—or, for that matter, any modality—is difficult to perform. Helical CT angiography is not immune to such obstacles. Notwithstanding the requirement for ionizing radiation, helical CT angiography, too, possesses modality-specific limitations. For example, the technique used by the Chopra et al requires the intravenous administration of iodinated contrast material, 150 mL injected at a rate of 5 mL/min. Two patients were excluded from the study because of inadequate venous access. Further, most patients and their physicians, when given the choice, would likely choose a modality that requires neither the placement of an intravenous line nor the intravenous administration of contrast material. That choice would be especially easy in patients with renal insufficiency.
The time required to perform helical CT angiography was addressed by the authors: They determined that data acquisition and postprocessing required, on average, 22 minutes. It is difficult to compare this period with that required to perform a US examination, as the latter was not recorded. Further, from a technical aspect, actual procedure time should include patient preparation time (eg, the time required to start an intravenous line) and room time. From a physician's standpoint, procedure time should include the time spent during image interpretation, as well. Each helical CT angiographic examination generated 66–122 transverse source images, many more than are obtained at US examination, and those images must be postprocessed to obtain MPR images. While Dr Chopra is experienced at such reformation, as was evidenced by the short image postprocessing time, there is likely to be an extensive learning curve, both in image reprocessing and in image interpretation, especially in centers in which only a small number of TIPS procedures are performed. That inexperience would likely result in much longer procedural times. The relative times required by each procedure, therefore, are unknown.
Finally, although cost was not addressed in the study (and need not have been, as this evaluation was directed toward diagnostic accuracy, not clinical utility), it is of great importance in clinical medicine. To my knowledge, an extensive cost-analysis of US and helical CT angiography in this setting is not available. However, costs for CT and US procedures of similar complexities suggest that US will likely prove to be cheaper than helical CT angiography in this setting.
It follows from the previous discussion that US and helical CT angiography are both acceptable modalities from a diagnostic standpoint. It is likely that the decision to use one or the other will be highly institutional-specific and dependent on local expertise. In addition, the choice of one procedure over the other should be patient-specific. One possible protocol might use US for the majority of patients while it reserves helical CT angiography (because of its requirement for ionizing radiation, intravenous access, and intravenous administration of contrast material) for those patients in whom US is not diagnostic or technically feasible.
I congratulate Dr Chopra and his colleagues for their development and initial evaluation of helical CT angiography as a potential screening modality for TIPS dysfunction. I wholeheartedly concur that further studies are necessary to determine the place of CT angiography in the clinical evaluation of malfunctioning TIPS. Because helical CT angiography allows evaluation of malfunctioning TIPS with a diagnostic accuracy similar to that of US, a prospective study in which the clinical utility of these modalities is compared, by using direct portography with manometry as the standard, is indicated. Helical CT angiography may prove to be a useful screening tool in the diagnosis of physiologically significant TIPS stenoses.
Footnotes
Abbreviations: MPR = multiplanar reformation TIPS = transjugular intrahepatic portosystemic shunt
See also the article by Chopra et al (pp 115–122 ) in this issue.
References
Related Article
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
