| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Letters to the Editor |
Mallinckrodt Institute of Radiology, Washington University School of Medicine, 510 South Kingshighway Boulevard, St Louis, MO 63110. e-mail: derdeync@wustl.edu
Editor:
I read with interest the cost-effectiveness analysis by Dr Buskens and colleagues in the October 2004 issue of Radiology. The authors discussed different imaging strategies for patients with recent cerebral or ocular symptoms and suspected stenosis of the carotid arterial bifurcation (1). Analyses such as these are critically dependent on (a) underlying assumptions of base-case risks and (b) the design of the mathematic model. With this in mind, I have several major reservations regarding the strength of the recommendations or conclusions that may be drawn from this generally well-executed and well-designed work.
The North American Symptomatic Carotid Endarterectomy Trial, or NASCET, and the European Carotid Surgery Trial, or ECST, were based on conventional angiography as the primary tool for the identification of surgical candidates (2,3). The risk for stroke with medical therapy was strongly related to the degree of stenosis, as determined by means of angiography, with the exception of patients with near-occlusion (4). While an ultrasonographic (US) or magnetic reso-nance (MR) angiographic examination is a reasonable screening tool for the identification of nonsignificant stenosis, if proved to be accurate by means of local quality assurance, the use of these tools to reliably identify patients with high-grade stenosis remains unclear. The trade-off is between the risks and costs of angiography versus the risks and costs of unnecessary endarterectomy (false-positive findings) and of a high stroke risk in patients that should have undergone endarterectomy (false-negative findings).
First, in the study by Dr Buskens and colleagues, the base-case risk of stroke with angiography was 3%. They state that at complication rates of less than 1%, little or nothing is gained by US or MR angiographic imaging strategies (after screening US). This result is critically important and is not emphasized in their discussion. Is a base-case assumption of a 3% risk for stroke valid for the hypothetical 55-year-old man used for this analysis? I think not. This rate comes from two studies with data largely compiled in the 1980s (5,6). Imaging equipment, catheters, and contrast agents have all improved since then. In addition, these studies certainly included many older patients at higher risk for stroke than a 55-year-old man.
Second, the incremental quality-adjusted life-years gained with the use of noninvasive imaging strategies are microscopic—11.2 to 11.3. Is this really meaningful?
Third, other imaging strategies have been proposed that are not analyzed here. Hathout et al (7) proposed the use of contrast material–enhanced MR angiography to screen patients with less than 50% stenosis and reliably identify patients with high-grade (greater than 80%) stenosis. Because of the wide confidence limits for MR angiography for the degree of stenosis in any given patient, conventional angiography was employed to address the 50%–80% stenosis group. This strategy could be employed for US, as well.
Fourth, the time horizon in this analysis is unusually long. The authors compiled quality-adjusted life years for the remaining life expectancy for a 55-year-old man. Most cost-effectiveness analyses are limited to 5-year time horizons. One reason for this is that the effects of other treatments that may become available are not included in the model. In addition, such a long horizon artificially magnifies nonsignificant differences in outcomes. Given the small differences in absolute quality-adjusted life-years between the different strategies here, this becomes a very real concern.
Fifth and finally, I was confused by their treatment of patients with complete carotid arterial occlusion. In table 3, the authors stated that complete occlusion was considered not to cause further ipsilateral events. The data are quite to the contrary. Patients with symptomatic complete carotid arterial occlusion are in fact at high risk for future ipsilateral stroke (8,9). A randomized trial of extra- to intracranial arterial bypass is underway in selected patients with symptomatic occlusion for this very reason.
In conclusion, the conclusions of this article should be tempered. Conventional angiography remains an important tool for the reliable identification of patients with surgically amenable carotid arterial stenosis, particularly at centers with documented complication rates of less than 1%.
REFERENCES
,
Yolanda van der Graaf, MD, PhD,* and
M. G. Myriam Hunink, MD, PhD
Julius Center for Health Sciences and Primary Care/HP Str. 6.131, University Medical Center Utrecht, PO Box 85500, Utrecht 3508 GA, the Netherlands*. e-mail: e.buskens@umcutrecht.nl
Department of Neurology, Academical Medical Center Amsterdam, the Netherlands
Department of Radiology, Erasmus Medical Center Rotterdam, the Netherlands
The close scrutiny of our article by Dr Derdeyn is highly appreciated. We believe we have addressed an issue that, despite a long history of carotid arterial imaging and subsequent surgery, still stirs the minds and maybe the odd heart. The article presented in Radiology is based on an original clinical study conducted in the Netherlands (1). To answer the question of which diagnostic strategy would allow the most accurate detection of severe (70%–99%) stenosis, 350 consecutive patients with neurologic symptoms who were referred for conventional angiography were invited to additionally undergo MR angiography subsequent to initial Doppler US. All relevant grades of stenosis were encountered in our sample. Moreover, the carotid arteries were assessed independently in a blinded fashion. Thus, we were able to obtain precise estimates of accuracy for all diagnostic tests over the entire range of stenoses.
We have, in fact, shown that noninvasive imaging indeed yields a precise measure of stenosis, with sensitivities and specifities up to 96% and 80% for MR angiography and Doppler US, respectively. Having noted this, the balance between costs and effects critically depends on the consequences of complications—that is, stroke, in both the short term and the long term. These may be associated with the diagnostic work-up itself, surgery, or progressive cardiovascular disease. As Dr Derdeyn correctly stated, high proportions of false-negative findings and/or false-positive findings might tip the balance. However, when diagnosing severe carotid arterial stenosis, a so-called false-positive result (relatively poor specificity) may still lead to a gain in (quality-adjusted) life expectancy, as patients with lower-grade stenosis may also benefit (2). With regard to false-negative findings, these rarely occur, especially if the threshold for Doppler US and/or MR angiography is chosen so as to optimize sensitivity.
With regard to the presumed high 3% estimate of complications arising from conventional angiography, we believe it is important to keep in mind the relevant domain. We are dealing with patients with advanced cardiovascular disorder and a potentially unstable plaque in the carotid arteries. A 55-year-old individual without atherosclerotic disease will probably have a lower risk of complications. However, selective catheterization of a carotid artery with loose plaque material and subsequent injection of contrast material carries considerable risk. In fact, in our own clinical study, we encountered a 2.3% complication rate—that is, stroke or death attributable solely to angiography (1).
Also, with regard to lower rates of neurologic complications cited from the literature, one has to critically assess the relevant domain. In fact, more recently, a lower rate of neurologic complications resulting from digital subtraction angiography was reported: 0.5% for stroke and 0.4% for transient ischemic attack (3). After closer inspection, however, it becomes apparent that for the subgroup of patients undergoing a diagnostic work-up for cardiovascular disease and/or suspected carotid arterial stenosis, high complication rates were reported—that is, over 2% and well above our 1% threshold. We chose to use a 3% risk for our base-case analysis, since, in our opinion, this figure still applies, even in the hands of experts, and is corroborated by sound evidence. Additionally, even patients without apparent neurologic complications after digital subtraction angiography have been shown to develop minor asymptomatic infarctions as a result of microembolisms, corroborating the real risk associated with conventional angiography (4).
As for the relatively small gain in quality-adjusted life-years, the readers should keep in mind that we are presenting expected outcome on the basis of a modeling study. This has no bearing on significance testing. Particularly, the fact that noninvasive imaging strategies consistently (and robustly) yielded better outcome while at the same time saving costs when compared with strategies such as conventional angiography provides solid evidence of superiority (ie, dominance for those familiar with the meaning in terms of economic evaluation) of noninvasive testing.
Subsequently, Dr Derdeyn suggests that other noninvasive techniques should possibly also have been included. This is a point well taken. We realize that our results should be considered a "cross-section" of the state of the art in the late 90s of the past century. We, for example, also added contrast-enhanced MR angiography to our diagnostic work-up, but because digital subtraction angiography is not performed routinely anymore, we could validate this technique only in a small series. Moreover, the diagnostic accuracy of contrast-enhanced MR angiography and, for example, computed tomographic angiography, have not yet been investigated in large prospective consecutive cohorts, and therefore, robust estimates for cost-effectiveness models remain unavailable (5,6).
Also, as already alluded to earlier, further improvement of test accuracy does not necessarily lead to more favorable cost-effectiveness. Especially, adding costly contrast agents to noninvasive tests such as MR angiography is likely to lead to less favorable incremental cost-effectiveness ratios, or iCERs. Indeed, very little may be gained, whereas the costs would increase considerably, resulting in astronomic iCERs.
The argument regarding the time horizon might also be turned around. Indeed, frequently, too short a time horizon is used. A short time horizon will lead to underestimation of the true gain in effectiveness and incremental costs that may be expected. The fact that new treatment options or diagnostic possibilities may become available in the future does not change the fact that the decision needs to be made now for the patients we care for now with the technologies we currently have available. Furthermore, potential unpredictable future developments do not release the researcher from aiming at a comprehensive evaluation in accordance with up-to-date insights and evidence.
Finally, the question is raised whether patients with complete occlusion are no longer considered at risk of ischemic stroke. This notion may have resulted from the complex nature of our model. We believe that the assumption that no further surgical treatment will be offered in the case of occlusion still holds in the context of the clinical setting that we have modeled. Thus, ipsilateral events associated with diagnosis, treatment, and further progression of the disease locally were not considered. However, by no means did we assume that future cerebrovascular events would no longer occur. In fact, as stated in the appendix to our report, we attributed an increased risk of future cerebrovascular and other cardiovascular events to the category of patients with carotid arterial occlusion.
In our opinion, previous reports on experimental carotid arterial bypass surgery have thus far not shown convincing evidence that overall health gain may be expected from surgical intervention for an occluded carotid artery. Important studies aimed at finding the best strategy for these patients have only recently started to enroll. If, in the future, evidence should become available that a surgical procedure also yields benefit for patients with occlusion, this may be taken into account. It appears unlikely, however, that this would substantially affect the cost-effectiveness of diagnostic strategies. Regardless of the extent of the stenosis, noninvasive strategies will avert the risk of complications, which no doubt remains high in the subgroup with occlusion.
In conclusion, we believe that we have convincingly and robustly shown that routine conventional angiography no longer has a place in daily practice and see no reason to temper our conclusions.
REFERENCES
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
