- Email: [email protected]
Quantification of iliac artery stenoses: a methodological comparative study between intravascular ultrasound, arteriography and duplex scanning
Ultrasound in Med. & Biol., Vol. 24, No. 7, pp. 963–970 Copyright © 1998 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/98 $19.00 1 .00
PII S0301-5629(98)00068-4
● Original Contribution QUANTIFICATION OF ILIAC ARTERY STENOSES: A METHODOLOGICAL COMPARATIVE STUDY BETWEEN INTRAVASCULAR ULTRASOUND, ARTERIOGRAPHY AND DUPLEX SCANNING KATJA C. VOGT,* JOHN G. RASMUSSEN,† LENE T. SKOVGAARD,‡ SVEN JUST† and TORBEN V. SCHROEDER* Departments of *Vascular Surgery and †Radiology, Section for Cardiovascular Radiology, and ‡Department of Biostatistics, Rigshospitalet, University of Copenhagen, Denmark (Received 3 December 1997; in final form 29 April 1998)
Abstract—Two morphological methods for quantifying the degree of stenoses in the iliac arteries, intravascular ultrasound (IVUS) and arteriography, were compared with duplex scanning, a method of evaluating the haemodynamic importance of the stenosis. A total of 38 patients, 20 women and 18 men, median age 66 y, admitted for either PTA (n 5 18) or femoro-femoral crossover bypass surgery (n 5 20), were examined by IVUS, single plane arteriography and duplex scanning. The predictive value, sensitivity, specificity and kappa value of IVUS were higher than the corresponding values for arteriography. Logistic regression analysis found that IVUS had a predictive value (p 5 0.0003) for diagnosing significant stenosis as defined by duplex scanning, but arteriography did not (p 5 0.1). However, this difference in usefulness as predictors did not reach significance. The agreement between arteriography and IVUS was relatively good. The conclusion is that IVUS seems to be superior to single-plane arteriography in predicting hemodynamically significant stenoses as defined by duplex scanning. © 1998 World Federation for Ultrasound in Medicine & Biology. Key Words: Intravascular ultrasound, Arteriography, Duplex scanning, Iliac artery stenoses, Atherosclerosis.
distinguished from the media by an echolucent layer (Di Mario et al. 1992). The precise degree of stenosis can, therefore, be calculated by direct measurement of the luminal area and the media-bounded area, with the limitation that areas of calcification produce shadowing due to the high echogenicity, and can prevent visualization of the whole circumference of the arterial wall (Di Mario et al. 1992; Gussenhoven et al. 1989; Nishimura et al. 1990). The method is invasive, but does not require X rays or contrast agents, and reports of complications in the literature are very rare. The catheter can be introduced both percutaneously and through an arteriotomy. It is, therefore, both easily interchangeable with the balloon catheter during PTA and can evaluate inflow problems during infrainguinal bypass surgery. The major drawback of the method is the cost of the disposable catheters. Since the first reports in the late 1980s, many authors have compared IVUS with the existing “gold standards,” namely histology (Di Mario et al. 1992; Gussenhoven et al. 1989; Nishimura et al. 1990) and arteriography (De Scheerder et al. 1994; Gerritsen et al. 1993;
INTRODUCTION Potential occlusive disease in the iliac arteries can be assessed by various methods, the easiest being palpation of femoral pulses. More detailed and precise information can be obtained from arteriography, which has been the “gold standard” for many years. However, the method is invasive and requires the use of contrast agents. Atherosclerotic plaques may be overlooked if they are eccentrically distributed, especially if the vessels are tortuous and not situated in the same plane from the aorta to the groin (Thiele and Strandness 1983). The advantage of arteriography is that it is easy to interpret and visualizes the whole arterial tree of the extremity. Intravascular ultrasound (IVUS) is a new imaging modality, particularly suitable for assessing endovascular procedures. It produces transsectional images of the artery at a high resolution (Fig. 1A, B). The intima can be Address correspondence to: Katja C. Vogt, Department of Vascular Surgery, RK3111, Rigshospitalet, University of Copenhagen, Blegdamsvej 9, DK-2100 Copenhagen, Denmark. E-mail: torben.s@ rh.dk 963
964
Ultrasound in Medicine and Biology
Volume 24, Number 7, 1998
raphy have been compared by several authors (Currie et al. 1995; De Smeet et al. 1996; Jaeger et al. 1985; Kohler et al. 1987; Langsfeld et al. 1988; Legemate et al. 1989, 1991; Moneta et al. 1992; Rosfors et al. 1993), who reported relatively good results but, to our knowledge, the agreement between IVUS and duplex scanning in untreated arteries has not previously been assessed. MATERIALS AND METHODS
Fig. 1. IVUS image of an iliac artery with an eccentric soft plaque. (A) The transducer (t), creates a black area in the center of the transsectional view. The lumen (l) is bordered by the intima, which is thickened by soft plaque (p) in the lower left part of the artery. The media (m) is represented by a black ring outside the intima. (B) The outer and inner lining of the intima have been marked. The area of the lumen and the media-bound area are calculated and shown in the lower left corner of the picture, along with the degree of stenosis and diameter reduction, which have been calculated from these two areas.
Tabbara et al. 1991). Both of these morphological methods have their limitations. Single-plane arteriography may, as mentioned, underestimate the degree of stenosis, and the necessary processing for histological examination causes disproportionate dimensional changes of the arteries (Di Mario et al. 1992; Siegel et al. 1985). The clinical importance of a given stenosis relates to its potential hemodynamic effect. We, therefore, wanted to compare IVUS and arteriography with duplex scanning in identifying hemodynamically significant lesions in the iliac arteries. Duplex scanning and arteriog-
A total of 38 patients taken from a series of 41 were admitted for either percutaneous transluminal angioplasty (PTA) of the iliac artery (n 5 18) or for femorofemoral crossover bypass (n 5 20) during the period from 1 June 1994 to 20 February 1997. The 3 patients had been excluded from the study because the time interval between arteriography and duplex examination exceeded 3 months. The median age of the 20 women and 18 men was 66 y (range: 44 to 82 y). Two (5%) patients suffered from diabetes. Of the total, 25 (66%) patients were current smokers and 9 (24%) patients had recently stopped smoking. The median ankle-brachial index (ABI) in the group of patients scheduled for PTA was 0.55 (0.31– 0.8), and the median ABI for the donor limb in the group of patients scheduled for crossover bypass was 0.68 (0.34 –1.0). The indication for PTA was claudication in 14 (78%) cases, rest pain in 3 (17%) cases, and 1 (5%) patient had gangrene. In the crossover group, 11 (55%) patients had no symptoms from the donor limb, 8 (40%) had claudication (2 of which were caused by infrainguinal arteriosclerosis) and 1 (5%) patient had nonhealing ulcers after radiation therapy. All patients were examined with IVUS, duplex scanning and arteriography prior to revascularization. The patients admitted for PTA had the diseased iliac artery examined, and the patients admitted for femoro-femoral crossover bypass had the donor iliac artery examined. Arteriography was performed just prior to the balloon inflation in the PTA patients, and the median time interval between arteriography and revascularization in the crossover patients was 18 days (0 days–3 months). The IVUS examination was performed just before and in connection with the revascularization in all cases. Finally, duplex scanning was attempted on the day before revascularization, with a median time interval of 1 (0–19) days. Six patients admitted for crossover bypass surgery underwent intraoperative PTA due to the discovery of significant stenosis in the donor iliac artery during the abovementioned examinations. The values used for this comparative study were the degree of stenosis measured before the PTA. Duplex scanning The patients were examined in the supine position using a B&K 3535 color duplex scanner (B&K Medical,
Quantification of iliac artery stenoses ● K. C. VOGT et al.
Copenhagen, Denmark) and a 3.5-MHz curved tranducer. Doppler spectra were obtained routinely from four sites: those were the proximal and distal segments of the common and the external iliac arteries. If a stenosis was present in between these locations, as indicated by the color-flow map, a Doppler spectrum from this site was also obtained. The peak systolic velocity (PSV), the rise time and the end diastolic velocity (EDV) from each site were noted, along with information on spectral broadening. A stenosis of 50% or more was defined as a ratio of more than 2 between the PSV in the stenosis and PSV in a normal part of the same arterial segment (Currie et al. 1995; Jaeger et al. 1985; Kohler et al. 1987; Langsfeld et al. 1988; Moneta et al. 1992; Rosfors et al. 1993).
965
sured on the retracted part of the catheter to define the location of the stenosis. The IVUS equipment enabled off-line measurement of the free lumen area and the media-bound area, and the ratio between these two areas yielded the degree of stenosis, see Fig. 1B. All measurements were performed by the same observer (K.C.V.), and repeated after a period of at least 1 month, to evaluate the reproducibility of the image assessment. The first of the two measurements was used when comparing the results with the other two methods. Statistics Agreement between two continuous variables was evaluated by the method decribed by Bland and Altmann (1986), in which the difference between two measurements is related to the average. The correlation between the averages and the differences was tested with Spearman rank correlation analysis. Kappa statistics were used for agreement betweeen the categorical data. Receiver operating characteristic (ROC) curves were used to visualize the sensitivity and specificity for different cut-off values for hemodynamically significant stenosis. The predictive values of IVUS and arteriography compared to duplex scanning were evaluated by logistic regression analysis. The identity of the regression coefficients when using both methods simultaneously in a logistic regression model was tested by a X2 test (22 logQ, Q being the likelihood ratio).
Arteriography Single-plane angiograms (anterior-posterior) were obtained after retrograde insertion of a sheath into the common femoral artery. The angiograms were reviewed by an experienced radiologist (J.G.R.), blinded to the duplex and IVUS results. The degree of stenosis was estimated by comparing the diameter in the lesion with the diameter in a seemingly normal part of the same artery. Additionally, an area reduction was calculated from the diameter reduction, assuming a circular lumen, to see if this could improve the comparison with the area reduction measured by IVUS. The runoff (i.e., patency of the superficial and deep femoral arteries), was noted. IVUS The intravascular ultrasound was performed with a 20-MHz transducer mounted in the tip of an 8 F catheter (CVIS Cardiovascular Imaging Systems Inc., Sunnyvale, CA). The catheter was introduced either through the arterial sheath used for the angioplasty in patients scheduled for PTA, or through an arteriotomy of the common femoral artery in patients undergoing femoro-femoral bypass surgery. The catheter was advanced retrogradely into the aorta, and then retracted slowly while producing transsectional images displayed on a monitor and stored on a S-VHS video recorder. The position of the transducer was recorded under flouroscopy in the PTA-patients. During surgery, the length from the bifurcation or from the internal iliac artery to a lesion site was mea-
RESULTS There were no immediate complications connected with the IVUS or duplex scanning. One (6%) patient suffered minor bleeding in the groin after the PTA, treated succesfully by compression. Of the total, 28 patients had both the deep and the superficial femoral artery patent, and 9 patients had occlusion or severe stenoses of the superficial femoral artery. In 1 patient, the runoff could not be evaluated on the arteriogram. The median and range of the values from the three methods are shown in Table 1. The degree of stenosis evaluated by arteriography is given both as diameter
Table 1. Median and (range) for the three different methods, according to patient category, PTA or crossover bypass
PTA Crossover All Patients
n
IVUS, % area reduction
Arteriography, % diameter reduction
Arteriography, % area reduction
Duplex scanning, PSV ratio
18 20 38
75 (58–99) 50 (22–78) 65 (22–99)
70 (35–99) 50 (0–80) 58 (0–99)
91 (58–99) 75 (0–96) 82 (0–99)
5.05 (1.4–9.15) 1.9 (1.15–6.3) 2.7 (1.15–9.15)
IVUS 5 intravascular ultrasound; PSV 5 peak systolic velocity.
966
Ultrasound in Medicine and Biology
Volume 24, Number 7, 1998
Fig. 2. ROC curves for IVUS (F), and arteriography (E) (diameter reduction: ——, area reduction: - - -, underlined numbers) showing the sensitivity and specificity for different cutoff values for hemodynamically significant stenoses. The kappa values for each cutoff value can be seen in Table 2.
reduction and as a calculated area reduction. Sensitivity and specificity and kappa values for different cutoff values for hemodynamically significant stenoses, defined by duplex scanning, were calculated. As can be seen in Fig. 2, the ROC-curve for IVUS is superior to both the curves for arteriography. The best agreement between IVUS and duplex scanning was obtained for a cutoff value of 55%, giving a sensitivity of 0.96, a specificity of 0.86, and a kappa value of 0.83. The optimal cutoff value for arteriography was 40 % for the diameter reduction, sensitivity 0.96, specificity 0.64 and kappa 0.64. For the area reduction the optimal cutoff value was 55% with a sensitivity of 1.00, a specificity of 0.64 and a kappa value of 0.69. As can be seen in Fig. 2, the two ROC curves for arteriographic diameter reduction and area reduction are identical; only the cutoff values are shifted parallel. The difference between the two methods, evaluated by logistic regression with duplex scanning as the dependent variable, is shown in Fig. 3A, B, where the slope of the IVUS curve is seen to be much steeper than the one for arteriography. The slopes were estimated to be 0.226 and 0.075, respectively, both of which were significantly greater than zero. If logistic regression was performed with duplex scanning as the dependent categorical variable, and both IVUS and arteriography as covariates at the same time, the coefficient of IVUS was significant (0.194, p 5 0.0003) but that of arteriography was not (0.045, p 5 0.1). The clinical relevance of this is that, when using IVUS for the examination of the iliac arteries, a supplementary arteriography will not add to
the predictive value. On the other hand, the IVUS information will improve the predictive value of arteriography. However, this difference in usefulness as predictors did not reach significance, because a test for identity of the two coefficients (likelihood ratio test) gives a X2 value of 3.73, with a corresponding p value of 0.054. The lack of significance is probably due to the limited number of data. The dotted curve on Fig. 3B represents the arteriographic degree of stenosis calculated as an area reduction. As can be seen, this alteration does not improve the predictive value of the arteriography (slope: 0.07, regression coefficient: 0.04, p 5 0.07). The agreement between the two morphological methods, arteriography and IVUS, is shown in Fig. 4A in a Bland & Altman (1986) plot. It can be seen that the values are not equally distributed around the mean difference (10%), but there is a tendency toward higher differences in the less pronounced stenoses. This tendency is confirmed by Spearman correlation analysis
Table 2. Kappa values for each cutoff value
IVUS Arteriography, diam. reduction Arteriography, area reduction
40%
50%
55%
60%
70%
0.33
0.69
0.83
0.67
0.47
0.64
0.59
0.36
0.42
0.45
0.61
0.61
0.69
0.64
0.59
Quantification of iliac artery stenoses ● K. C. VOGT et al.
967
Fig. 3. Logistic regression analysis of (A) IVUS and (B) arteriography (diameter reduction: ——; area reduction: - - -) plotted against the duplex scanning as the dependent variable. 1 indicates significant stenosis (PSV ratio . 2.0) and 0 indicates no significant stenosis (PSV , 2.0).
(r 5 0.37, p 5 0.02), indicating a systematic discrepancy between the two methods. The observed differences ranged from 222 to 160% stenoses, with a standard deviation of 21%. Transforming the arteriographic degree of stenosis to an area reduction, Fig. 4B, diminishes the difference between the two methods (mean differ-
ence 26% (range 250 to 160, SD 27%), and the correlation between the difference and the average can no longer be found (r 5 0.27, p 5 0.1). The reproducibility of IVUS was excellent with a mean difference (6 2 SD) between the two measurements of 22% (6 7.9%); see Fig. 5.
968
Ultrasound in Medicine and Biology
Volume 24, Number 7, 1998
Fig. 4. Bland and Altman (1986) plots, where the average between IVUS and arteriography for each stenosis is plotted against the difference. (A) IVUS and arteriographic diameter reduction. Mean: 10 6 42%. There is a significant correlation between the averages and the differences (r 5 0.37, p 5 0.02). (B) IVUS and arteriographic area reduction. Mean difference: 26 6 52%. There is no significant correlation between the averages and the differences (r 5 0.27, p 5 0.1).
DISCUSSION Better agreement was found between duplex scanning and IVUS than between duplex scanning and arteriography. This was confirmed by kappa evaluation, ROC curves (Fig. 2) and logistic regression analysis (Fig. 3). The lack of predictive value of arteriography assessed against duplex may be related to the fact that only single-plane investigation was performed, which may cause eccentric plaques in the arteries to be missed (Thiele and Strandness 1983), thereby underestimating the less-pronounced stenoses. In addition, comparing the lumen in the stenosis to the lumen in part of the artery
that could easily be atherosclerotically diseased, will cause an underestimation of the degree of stenosis. It is, therefore, not surprising that the optimal cutoff value for arteriography, as evaluated from both the ROC and logistic regression analysis, was around 40 – 45%. On the other hand, IVUS, by calculating the degree of stenosis relating the luminal area to the media-bound area (Gerritsen et al. 1993; Gussenhoven et al. 1989; Jain et al. 1994; Tabbara et al. 1991; Vogt et al. 1997), will cause an overestimation due to the inclusion of the normal part of the intima. This explains the higher optimal cutoff value of 55– 60% for IVUS in discriminating hemody-
Quantification of iliac artery stenoses ● K. C. VOGT et al.
969
Fig. 5. Bland and Altman (1986) plot showing the reproducibility of IVUS.
namically significant stenoses as evaluated from the ROC curve, kappa values, and logistic regression analysis. In comparing the two methods, it is notable that, for all but one cutoff value (Fig. 2), the sensitivity and specificity of IVUS were superior to arteriography when using duplex scanning to define the hemodynamically significant stenoses. The different principles for calculating the degree of stenosis could explain the systematic difference between IVUS and arteriography, particularly concerning the moderate lesions (Fig. 4A). Due to the above-mentioned differences between arteriography and IVUS, measuring the degree of stenosis as a diameter reduction and, hence, an area reduction, we calculated an area reduction for the arteriography from the diameter reduction value. Because the inherent limitations of the single-plane arteriography are not eliminated by this transformation, the comparison to duplex scanning, accuracy and predictive value, are not improved; the ROC curve and the slope of the regression curve are unchanged. However, when comparing the degree of stenosis directly with IVUS in the Bland and Altmann (1986) plot, Fig. 4B, a discrete improvement in
the agreement can be seen, and the systematic difference is now eliminated. When duplex scanning was introduced as a possible method for evaluating atherosclerotic disease in the iliac vessels, it was compared to arteriography. Different PSV ratios have been suggested optimally to distinguish whether a stenosis was hemodynamically significant or not. Most authors agree on a ratio of 2 or 2.5 (Currie et al. 1995; De Smeet et al. 1996; Jaeger et al. 1985; Kohler et al. 1987; Langsfeld et al. 1988; Legemate et al. 1989, 1991; Moneta et al. 1992; Rosfors et al. 1993). Using these values, sensitivities between 0.82 to 0.91 and specificities from 0.92 to 1.0 have been reported. However, if arteries with minor lesions were also included, as in our study, the agreement deteriorated considerably (Langsfeld et al. 1988). We routinely perform only single-plane arteriographies, which may also explain our results of a sensitivity of 0.92 but a specificity of only 0.66. As mentioned, the disagreement between IVUS and arteriography was largest in the less-pronounced stenoses, which were almost exclusively detected in the donor arteries of the crossover candidates. This suggests that
970
Ultrasound in Medicine and Biology
IVUS might be implemented during surgery, if visual evaluation or pressure measurement suggest an inflow problem that had not been detected on the preoperative arteriogram. If PTA was deemed necessary, the IVUS could also evaluate the result of this additional intervention, avoiding the use of fluoroscopy and contrast agents (Vogt et al. 1997). In the more severely diseased iliac arteries found in patients submitted for PTA, the agreement between the two methods, IVUS and arteriography, was better. However, other studies have shown that this agreement diminishes considerably after PTA (De Scheerder et al. 1994), where IVUS also seems to be superior in evaluating the morphological outcome of the intervention (Jain et al. 1994; Vogt et al. 1997). Therefore, PTA of iliac or femoral arteries directed solely by IVUS, avoiding contrast and obtaining a better appreciation of dissection or residual stenosis, would seem an attractive future application. We conclude that IVUS seems to be superior to single-plane arteriography in predicting hemodynamically significant stenoses defined by duplex scanning, although, in this small patient group, the difference did not reach a statistically significant level. Acknowledgements—This study was supported by the Velux Foundation and by Minister Erna Hamilton’s Foundation for Science and Art.
REFERENCES Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307– 310. Currie IC, Jones AJ, Wakeley CJ, et al. Non-invasive aortoiliac assessment. Eur J Vasc Endovasc Surg 1995;9:24 –28. De Scheerder I, De Man F, Herregods MC, et al. Intravascular ultrasound versus angiography for measurement of luminal diameters in normal and diseased coronary arteries. Am Heart J 1994;127:243– 251. De Smeet AAEA, Ermers EJM, Kitslaar PJEHM. Duplex velocity characteristics of aortoiliac stenoses. J Vasc Surg 1996;23:628 – 636.
Volume 24, Number 7, 1998 Di Mario C,The SHK, Madretsma S, et al. Detection and characterization of vascular lesions by intravascular ultrasound: An in vitro study correlated with histology. J Am Soc Echocardiogr 1992;5: 135–146. Gerritsen GP, Gussenhoven EJ, The SHK, et al. Intravascular ultrasound before and after intervention: In vivo comparison with angiography. J Vasc Surg 1993;18:31– 40. Gussenhoven EJ, Essed CE, Frietman P, et al. Intravascular ultrasonic imaging: histological and echographic correlation. Eur J Vasc Surg 1989;3:571–576. Jaeger KA, Phillips DJ, Martin RL. Non-invasive mapping of lower limb arterial lesions. Ultrasound Med Biol 1985;11:515–521. Jain SP, Jain A, Collins TJ, Ramee SR, White CJ. Predictors of restenosis: A morphometric and quantitative evaluation by intravascular ultrasound. Am Heart J. 1994;128:664 – 673. Kohler TR, Nance DR, Cramer MM, Vanderburghe N, Strandness DE. Duplex scanning for the diagnosis of aorto-iliac and femorpopliteal disease: a prospective study. Circulation 1987;76:1074 –1080. Langsfeld M, Nepute J, Hershey FB, et al. The use of deep duplex scanning to predict hemodynamically significant aortoiliac stenoses. J Vasc Surg 1988;7:395–399. Legemate DA, Teeuwen C, Hoeneveld H, Eikelboom BC. The potential of duplex scanning to replace aorto-iliac and femoropopliteal angiography. Eur J Vasc Surg 1989;3:49 –54. Legemate DA, Teeuwen C, Hoeneveld H, Eikelboom BC. Value of duplex scanning compared with angiography and pressure measurement in the assessment of aortoiliac arterial lesions. Br J Surg 1991;78:1003–1008. Moneta GL, Yeager RA, Antonovic R, et al. Accuracy of lower limb extremity arterial duplex mapping. J Vasc Surg 1992;15:275–284. Nishimura RA, Edwards WD, Warnes CA, et al. Intravascular ultrasound imaging: in vitro validation and pathologic correlation. J Am Coll Cardiol 1990;16:145–154. Rosfors S, Eriksson M, Hglund N, Johansson G. Duplex ultrasound in patients with suspected aorto-iliac occlusive disease. Eur J Vasc Surg 1993;7:513–517. Siegel RJ, Swan K, Edwards G, Fishbein MC. Limitations of postmortem assessment of human coronary artery size and luminal narrowing: differential effects of tissue fixation and processing on vessels with different degrees of atherosclerosis. J Am Coll Cardiol 1985; 5:342–346. Tabbara M, White R, Cavaye D, Kopchok G. In vivo comparison of intravascular ultrasound and angiography. J Vasc Surg 1991;14: 496 –504. Thiele BL, Strandness DE Jr. Accuracy of angiographic quantification of peripheral atherosclerosis. Prog Cardiovasc Dis 1983;26:223– 236. Vogt KC, Just S, Rasmussen JG, Schroeder TV. Prediction of outcome after femoropopliteal balloon angioplasty by intravascular ultrasound. Eur J Vasc Endovasc Surg 1997;13:563–568.
PII S0301-5629(98)00068-4
● Original Contribution QUANTIFICATION OF ILIAC ARTERY STENOSES: A METHODOLOGICAL COMPARATIVE STUDY BETWEEN INTRAVASCULAR ULTRASOUND, ARTERIOGRAPHY AND DUPLEX SCANNING KATJA C. VOGT,* JOHN G. RASMUSSEN,† LENE T. SKOVGAARD,‡ SVEN JUST† and TORBEN V. SCHROEDER* Departments of *Vascular Surgery and †Radiology, Section for Cardiovascular Radiology, and ‡Department of Biostatistics, Rigshospitalet, University of Copenhagen, Denmark (Received 3 December 1997; in final form 29 April 1998)
Abstract—Two morphological methods for quantifying the degree of stenoses in the iliac arteries, intravascular ultrasound (IVUS) and arteriography, were compared with duplex scanning, a method of evaluating the haemodynamic importance of the stenosis. A total of 38 patients, 20 women and 18 men, median age 66 y, admitted for either PTA (n 5 18) or femoro-femoral crossover bypass surgery (n 5 20), were examined by IVUS, single plane arteriography and duplex scanning. The predictive value, sensitivity, specificity and kappa value of IVUS were higher than the corresponding values for arteriography. Logistic regression analysis found that IVUS had a predictive value (p 5 0.0003) for diagnosing significant stenosis as defined by duplex scanning, but arteriography did not (p 5 0.1). However, this difference in usefulness as predictors did not reach significance. The agreement between arteriography and IVUS was relatively good. The conclusion is that IVUS seems to be superior to single-plane arteriography in predicting hemodynamically significant stenoses as defined by duplex scanning. © 1998 World Federation for Ultrasound in Medicine & Biology. Key Words: Intravascular ultrasound, Arteriography, Duplex scanning, Iliac artery stenoses, Atherosclerosis.
distinguished from the media by an echolucent layer (Di Mario et al. 1992). The precise degree of stenosis can, therefore, be calculated by direct measurement of the luminal area and the media-bounded area, with the limitation that areas of calcification produce shadowing due to the high echogenicity, and can prevent visualization of the whole circumference of the arterial wall (Di Mario et al. 1992; Gussenhoven et al. 1989; Nishimura et al. 1990). The method is invasive, but does not require X rays or contrast agents, and reports of complications in the literature are very rare. The catheter can be introduced both percutaneously and through an arteriotomy. It is, therefore, both easily interchangeable with the balloon catheter during PTA and can evaluate inflow problems during infrainguinal bypass surgery. The major drawback of the method is the cost of the disposable catheters. Since the first reports in the late 1980s, many authors have compared IVUS with the existing “gold standards,” namely histology (Di Mario et al. 1992; Gussenhoven et al. 1989; Nishimura et al. 1990) and arteriography (De Scheerder et al. 1994; Gerritsen et al. 1993;
INTRODUCTION Potential occlusive disease in the iliac arteries can be assessed by various methods, the easiest being palpation of femoral pulses. More detailed and precise information can be obtained from arteriography, which has been the “gold standard” for many years. However, the method is invasive and requires the use of contrast agents. Atherosclerotic plaques may be overlooked if they are eccentrically distributed, especially if the vessels are tortuous and not situated in the same plane from the aorta to the groin (Thiele and Strandness 1983). The advantage of arteriography is that it is easy to interpret and visualizes the whole arterial tree of the extremity. Intravascular ultrasound (IVUS) is a new imaging modality, particularly suitable for assessing endovascular procedures. It produces transsectional images of the artery at a high resolution (Fig. 1A, B). The intima can be Address correspondence to: Katja C. Vogt, Department of Vascular Surgery, RK3111, Rigshospitalet, University of Copenhagen, Blegdamsvej 9, DK-2100 Copenhagen, Denmark. E-mail: torben.s@ rh.dk 963
964
Ultrasound in Medicine and Biology
Volume 24, Number 7, 1998
raphy have been compared by several authors (Currie et al. 1995; De Smeet et al. 1996; Jaeger et al. 1985; Kohler et al. 1987; Langsfeld et al. 1988; Legemate et al. 1989, 1991; Moneta et al. 1992; Rosfors et al. 1993), who reported relatively good results but, to our knowledge, the agreement between IVUS and duplex scanning in untreated arteries has not previously been assessed. MATERIALS AND METHODS
Fig. 1. IVUS image of an iliac artery with an eccentric soft plaque. (A) The transducer (t), creates a black area in the center of the transsectional view. The lumen (l) is bordered by the intima, which is thickened by soft plaque (p) in the lower left part of the artery. The media (m) is represented by a black ring outside the intima. (B) The outer and inner lining of the intima have been marked. The area of the lumen and the media-bound area are calculated and shown in the lower left corner of the picture, along with the degree of stenosis and diameter reduction, which have been calculated from these two areas.
Tabbara et al. 1991). Both of these morphological methods have their limitations. Single-plane arteriography may, as mentioned, underestimate the degree of stenosis, and the necessary processing for histological examination causes disproportionate dimensional changes of the arteries (Di Mario et al. 1992; Siegel et al. 1985). The clinical importance of a given stenosis relates to its potential hemodynamic effect. We, therefore, wanted to compare IVUS and arteriography with duplex scanning in identifying hemodynamically significant lesions in the iliac arteries. Duplex scanning and arteriog-
A total of 38 patients taken from a series of 41 were admitted for either percutaneous transluminal angioplasty (PTA) of the iliac artery (n 5 18) or for femorofemoral crossover bypass (n 5 20) during the period from 1 June 1994 to 20 February 1997. The 3 patients had been excluded from the study because the time interval between arteriography and duplex examination exceeded 3 months. The median age of the 20 women and 18 men was 66 y (range: 44 to 82 y). Two (5%) patients suffered from diabetes. Of the total, 25 (66%) patients were current smokers and 9 (24%) patients had recently stopped smoking. The median ankle-brachial index (ABI) in the group of patients scheduled for PTA was 0.55 (0.31– 0.8), and the median ABI for the donor limb in the group of patients scheduled for crossover bypass was 0.68 (0.34 –1.0). The indication for PTA was claudication in 14 (78%) cases, rest pain in 3 (17%) cases, and 1 (5%) patient had gangrene. In the crossover group, 11 (55%) patients had no symptoms from the donor limb, 8 (40%) had claudication (2 of which were caused by infrainguinal arteriosclerosis) and 1 (5%) patient had nonhealing ulcers after radiation therapy. All patients were examined with IVUS, duplex scanning and arteriography prior to revascularization. The patients admitted for PTA had the diseased iliac artery examined, and the patients admitted for femoro-femoral crossover bypass had the donor iliac artery examined. Arteriography was performed just prior to the balloon inflation in the PTA patients, and the median time interval between arteriography and revascularization in the crossover patients was 18 days (0 days–3 months). The IVUS examination was performed just before and in connection with the revascularization in all cases. Finally, duplex scanning was attempted on the day before revascularization, with a median time interval of 1 (0–19) days. Six patients admitted for crossover bypass surgery underwent intraoperative PTA due to the discovery of significant stenosis in the donor iliac artery during the abovementioned examinations. The values used for this comparative study were the degree of stenosis measured before the PTA. Duplex scanning The patients were examined in the supine position using a B&K 3535 color duplex scanner (B&K Medical,
Quantification of iliac artery stenoses ● K. C. VOGT et al.
Copenhagen, Denmark) and a 3.5-MHz curved tranducer. Doppler spectra were obtained routinely from four sites: those were the proximal and distal segments of the common and the external iliac arteries. If a stenosis was present in between these locations, as indicated by the color-flow map, a Doppler spectrum from this site was also obtained. The peak systolic velocity (PSV), the rise time and the end diastolic velocity (EDV) from each site were noted, along with information on spectral broadening. A stenosis of 50% or more was defined as a ratio of more than 2 between the PSV in the stenosis and PSV in a normal part of the same arterial segment (Currie et al. 1995; Jaeger et al. 1985; Kohler et al. 1987; Langsfeld et al. 1988; Moneta et al. 1992; Rosfors et al. 1993).
965
sured on the retracted part of the catheter to define the location of the stenosis. The IVUS equipment enabled off-line measurement of the free lumen area and the media-bound area, and the ratio between these two areas yielded the degree of stenosis, see Fig. 1B. All measurements were performed by the same observer (K.C.V.), and repeated after a period of at least 1 month, to evaluate the reproducibility of the image assessment. The first of the two measurements was used when comparing the results with the other two methods. Statistics Agreement between two continuous variables was evaluated by the method decribed by Bland and Altmann (1986), in which the difference between two measurements is related to the average. The correlation between the averages and the differences was tested with Spearman rank correlation analysis. Kappa statistics were used for agreement betweeen the categorical data. Receiver operating characteristic (ROC) curves were used to visualize the sensitivity and specificity for different cut-off values for hemodynamically significant stenosis. The predictive values of IVUS and arteriography compared to duplex scanning were evaluated by logistic regression analysis. The identity of the regression coefficients when using both methods simultaneously in a logistic regression model was tested by a X2 test (22 logQ, Q being the likelihood ratio).
Arteriography Single-plane angiograms (anterior-posterior) were obtained after retrograde insertion of a sheath into the common femoral artery. The angiograms were reviewed by an experienced radiologist (J.G.R.), blinded to the duplex and IVUS results. The degree of stenosis was estimated by comparing the diameter in the lesion with the diameter in a seemingly normal part of the same artery. Additionally, an area reduction was calculated from the diameter reduction, assuming a circular lumen, to see if this could improve the comparison with the area reduction measured by IVUS. The runoff (i.e., patency of the superficial and deep femoral arteries), was noted. IVUS The intravascular ultrasound was performed with a 20-MHz transducer mounted in the tip of an 8 F catheter (CVIS Cardiovascular Imaging Systems Inc., Sunnyvale, CA). The catheter was introduced either through the arterial sheath used for the angioplasty in patients scheduled for PTA, or through an arteriotomy of the common femoral artery in patients undergoing femoro-femoral bypass surgery. The catheter was advanced retrogradely into the aorta, and then retracted slowly while producing transsectional images displayed on a monitor and stored on a S-VHS video recorder. The position of the transducer was recorded under flouroscopy in the PTA-patients. During surgery, the length from the bifurcation or from the internal iliac artery to a lesion site was mea-
RESULTS There were no immediate complications connected with the IVUS or duplex scanning. One (6%) patient suffered minor bleeding in the groin after the PTA, treated succesfully by compression. Of the total, 28 patients had both the deep and the superficial femoral artery patent, and 9 patients had occlusion or severe stenoses of the superficial femoral artery. In 1 patient, the runoff could not be evaluated on the arteriogram. The median and range of the values from the three methods are shown in Table 1. The degree of stenosis evaluated by arteriography is given both as diameter
Table 1. Median and (range) for the three different methods, according to patient category, PTA or crossover bypass
PTA Crossover All Patients
n
IVUS, % area reduction
Arteriography, % diameter reduction
Arteriography, % area reduction
Duplex scanning, PSV ratio
18 20 38
75 (58–99) 50 (22–78) 65 (22–99)
70 (35–99) 50 (0–80) 58 (0–99)
91 (58–99) 75 (0–96) 82 (0–99)
5.05 (1.4–9.15) 1.9 (1.15–6.3) 2.7 (1.15–9.15)
IVUS 5 intravascular ultrasound; PSV 5 peak systolic velocity.
966
Ultrasound in Medicine and Biology
Volume 24, Number 7, 1998
Fig. 2. ROC curves for IVUS (F), and arteriography (E) (diameter reduction: ——, area reduction: - - -, underlined numbers) showing the sensitivity and specificity for different cutoff values for hemodynamically significant stenoses. The kappa values for each cutoff value can be seen in Table 2.
reduction and as a calculated area reduction. Sensitivity and specificity and kappa values for different cutoff values for hemodynamically significant stenoses, defined by duplex scanning, were calculated. As can be seen in Fig. 2, the ROC-curve for IVUS is superior to both the curves for arteriography. The best agreement between IVUS and duplex scanning was obtained for a cutoff value of 55%, giving a sensitivity of 0.96, a specificity of 0.86, and a kappa value of 0.83. The optimal cutoff value for arteriography was 40 % for the diameter reduction, sensitivity 0.96, specificity 0.64 and kappa 0.64. For the area reduction the optimal cutoff value was 55% with a sensitivity of 1.00, a specificity of 0.64 and a kappa value of 0.69. As can be seen in Fig. 2, the two ROC curves for arteriographic diameter reduction and area reduction are identical; only the cutoff values are shifted parallel. The difference between the two methods, evaluated by logistic regression with duplex scanning as the dependent variable, is shown in Fig. 3A, B, where the slope of the IVUS curve is seen to be much steeper than the one for arteriography. The slopes were estimated to be 0.226 and 0.075, respectively, both of which were significantly greater than zero. If logistic regression was performed with duplex scanning as the dependent categorical variable, and both IVUS and arteriography as covariates at the same time, the coefficient of IVUS was significant (0.194, p 5 0.0003) but that of arteriography was not (0.045, p 5 0.1). The clinical relevance of this is that, when using IVUS for the examination of the iliac arteries, a supplementary arteriography will not add to
the predictive value. On the other hand, the IVUS information will improve the predictive value of arteriography. However, this difference in usefulness as predictors did not reach significance, because a test for identity of the two coefficients (likelihood ratio test) gives a X2 value of 3.73, with a corresponding p value of 0.054. The lack of significance is probably due to the limited number of data. The dotted curve on Fig. 3B represents the arteriographic degree of stenosis calculated as an area reduction. As can be seen, this alteration does not improve the predictive value of the arteriography (slope: 0.07, regression coefficient: 0.04, p 5 0.07). The agreement between the two morphological methods, arteriography and IVUS, is shown in Fig. 4A in a Bland & Altman (1986) plot. It can be seen that the values are not equally distributed around the mean difference (10%), but there is a tendency toward higher differences in the less pronounced stenoses. This tendency is confirmed by Spearman correlation analysis
Table 2. Kappa values for each cutoff value
IVUS Arteriography, diam. reduction Arteriography, area reduction
40%
50%
55%
60%
70%
0.33
0.69
0.83
0.67
0.47
0.64
0.59
0.36
0.42
0.45
0.61
0.61
0.69
0.64
0.59
Quantification of iliac artery stenoses ● K. C. VOGT et al.
967
Fig. 3. Logistic regression analysis of (A) IVUS and (B) arteriography (diameter reduction: ——; area reduction: - - -) plotted against the duplex scanning as the dependent variable. 1 indicates significant stenosis (PSV ratio . 2.0) and 0 indicates no significant stenosis (PSV , 2.0).
(r 5 0.37, p 5 0.02), indicating a systematic discrepancy between the two methods. The observed differences ranged from 222 to 160% stenoses, with a standard deviation of 21%. Transforming the arteriographic degree of stenosis to an area reduction, Fig. 4B, diminishes the difference between the two methods (mean differ-
ence 26% (range 250 to 160, SD 27%), and the correlation between the difference and the average can no longer be found (r 5 0.27, p 5 0.1). The reproducibility of IVUS was excellent with a mean difference (6 2 SD) between the two measurements of 22% (6 7.9%); see Fig. 5.
968
Ultrasound in Medicine and Biology
Volume 24, Number 7, 1998
Fig. 4. Bland and Altman (1986) plots, where the average between IVUS and arteriography for each stenosis is plotted against the difference. (A) IVUS and arteriographic diameter reduction. Mean: 10 6 42%. There is a significant correlation between the averages and the differences (r 5 0.37, p 5 0.02). (B) IVUS and arteriographic area reduction. Mean difference: 26 6 52%. There is no significant correlation between the averages and the differences (r 5 0.27, p 5 0.1).
DISCUSSION Better agreement was found between duplex scanning and IVUS than between duplex scanning and arteriography. This was confirmed by kappa evaluation, ROC curves (Fig. 2) and logistic regression analysis (Fig. 3). The lack of predictive value of arteriography assessed against duplex may be related to the fact that only single-plane investigation was performed, which may cause eccentric plaques in the arteries to be missed (Thiele and Strandness 1983), thereby underestimating the less-pronounced stenoses. In addition, comparing the lumen in the stenosis to the lumen in part of the artery
that could easily be atherosclerotically diseased, will cause an underestimation of the degree of stenosis. It is, therefore, not surprising that the optimal cutoff value for arteriography, as evaluated from both the ROC and logistic regression analysis, was around 40 – 45%. On the other hand, IVUS, by calculating the degree of stenosis relating the luminal area to the media-bound area (Gerritsen et al. 1993; Gussenhoven et al. 1989; Jain et al. 1994; Tabbara et al. 1991; Vogt et al. 1997), will cause an overestimation due to the inclusion of the normal part of the intima. This explains the higher optimal cutoff value of 55– 60% for IVUS in discriminating hemody-
Quantification of iliac artery stenoses ● K. C. VOGT et al.
969
Fig. 5. Bland and Altman (1986) plot showing the reproducibility of IVUS.
namically significant stenoses as evaluated from the ROC curve, kappa values, and logistic regression analysis. In comparing the two methods, it is notable that, for all but one cutoff value (Fig. 2), the sensitivity and specificity of IVUS were superior to arteriography when using duplex scanning to define the hemodynamically significant stenoses. The different principles for calculating the degree of stenosis could explain the systematic difference between IVUS and arteriography, particularly concerning the moderate lesions (Fig. 4A). Due to the above-mentioned differences between arteriography and IVUS, measuring the degree of stenosis as a diameter reduction and, hence, an area reduction, we calculated an area reduction for the arteriography from the diameter reduction value. Because the inherent limitations of the single-plane arteriography are not eliminated by this transformation, the comparison to duplex scanning, accuracy and predictive value, are not improved; the ROC curve and the slope of the regression curve are unchanged. However, when comparing the degree of stenosis directly with IVUS in the Bland and Altmann (1986) plot, Fig. 4B, a discrete improvement in
the agreement can be seen, and the systematic difference is now eliminated. When duplex scanning was introduced as a possible method for evaluating atherosclerotic disease in the iliac vessels, it was compared to arteriography. Different PSV ratios have been suggested optimally to distinguish whether a stenosis was hemodynamically significant or not. Most authors agree on a ratio of 2 or 2.5 (Currie et al. 1995; De Smeet et al. 1996; Jaeger et al. 1985; Kohler et al. 1987; Langsfeld et al. 1988; Legemate et al. 1989, 1991; Moneta et al. 1992; Rosfors et al. 1993). Using these values, sensitivities between 0.82 to 0.91 and specificities from 0.92 to 1.0 have been reported. However, if arteries with minor lesions were also included, as in our study, the agreement deteriorated considerably (Langsfeld et al. 1988). We routinely perform only single-plane arteriographies, which may also explain our results of a sensitivity of 0.92 but a specificity of only 0.66. As mentioned, the disagreement between IVUS and arteriography was largest in the less-pronounced stenoses, which were almost exclusively detected in the donor arteries of the crossover candidates. This suggests that
970
Ultrasound in Medicine and Biology
IVUS might be implemented during surgery, if visual evaluation or pressure measurement suggest an inflow problem that had not been detected on the preoperative arteriogram. If PTA was deemed necessary, the IVUS could also evaluate the result of this additional intervention, avoiding the use of fluoroscopy and contrast agents (Vogt et al. 1997). In the more severely diseased iliac arteries found in patients submitted for PTA, the agreement between the two methods, IVUS and arteriography, was better. However, other studies have shown that this agreement diminishes considerably after PTA (De Scheerder et al. 1994), where IVUS also seems to be superior in evaluating the morphological outcome of the intervention (Jain et al. 1994; Vogt et al. 1997). Therefore, PTA of iliac or femoral arteries directed solely by IVUS, avoiding contrast and obtaining a better appreciation of dissection or residual stenosis, would seem an attractive future application. We conclude that IVUS seems to be superior to single-plane arteriography in predicting hemodynamically significant stenoses defined by duplex scanning, although, in this small patient group, the difference did not reach a statistically significant level. Acknowledgements—This study was supported by the Velux Foundation and by Minister Erna Hamilton’s Foundation for Science and Art.
REFERENCES Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;1:307– 310. Currie IC, Jones AJ, Wakeley CJ, et al. Non-invasive aortoiliac assessment. Eur J Vasc Endovasc Surg 1995;9:24 –28. De Scheerder I, De Man F, Herregods MC, et al. Intravascular ultrasound versus angiography for measurement of luminal diameters in normal and diseased coronary arteries. Am Heart J 1994;127:243– 251. De Smeet AAEA, Ermers EJM, Kitslaar PJEHM. Duplex velocity characteristics of aortoiliac stenoses. J Vasc Surg 1996;23:628 – 636.
Volume 24, Number 7, 1998 Di Mario C,The SHK, Madretsma S, et al. Detection and characterization of vascular lesions by intravascular ultrasound: An in vitro study correlated with histology. J Am Soc Echocardiogr 1992;5: 135–146. Gerritsen GP, Gussenhoven EJ, The SHK, et al. Intravascular ultrasound before and after intervention: In vivo comparison with angiography. J Vasc Surg 1993;18:31– 40. Gussenhoven EJ, Essed CE, Frietman P, et al. Intravascular ultrasonic imaging: histological and echographic correlation. Eur J Vasc Surg 1989;3:571–576. Jaeger KA, Phillips DJ, Martin RL. Non-invasive mapping of lower limb arterial lesions. Ultrasound Med Biol 1985;11:515–521. Jain SP, Jain A, Collins TJ, Ramee SR, White CJ. Predictors of restenosis: A morphometric and quantitative evaluation by intravascular ultrasound. Am Heart J. 1994;128:664 – 673. Kohler TR, Nance DR, Cramer MM, Vanderburghe N, Strandness DE. Duplex scanning for the diagnosis of aorto-iliac and femorpopliteal disease: a prospective study. Circulation 1987;76:1074 –1080. Langsfeld M, Nepute J, Hershey FB, et al. The use of deep duplex scanning to predict hemodynamically significant aortoiliac stenoses. J Vasc Surg 1988;7:395–399. Legemate DA, Teeuwen C, Hoeneveld H, Eikelboom BC. The potential of duplex scanning to replace aorto-iliac and femoropopliteal angiography. Eur J Vasc Surg 1989;3:49 –54. Legemate DA, Teeuwen C, Hoeneveld H, Eikelboom BC. Value of duplex scanning compared with angiography and pressure measurement in the assessment of aortoiliac arterial lesions. Br J Surg 1991;78:1003–1008. Moneta GL, Yeager RA, Antonovic R, et al. Accuracy of lower limb extremity arterial duplex mapping. J Vasc Surg 1992;15:275–284. Nishimura RA, Edwards WD, Warnes CA, et al. Intravascular ultrasound imaging: in vitro validation and pathologic correlation. J Am Coll Cardiol 1990;16:145–154. Rosfors S, Eriksson M, Hglund N, Johansson G. Duplex ultrasound in patients with suspected aorto-iliac occlusive disease. Eur J Vasc Surg 1993;7:513–517. Siegel RJ, Swan K, Edwards G, Fishbein MC. Limitations of postmortem assessment of human coronary artery size and luminal narrowing: differential effects of tissue fixation and processing on vessels with different degrees of atherosclerosis. J Am Coll Cardiol 1985; 5:342–346. Tabbara M, White R, Cavaye D, Kopchok G. In vivo comparison of intravascular ultrasound and angiography. J Vasc Surg 1991;14: 496 –504. Thiele BL, Strandness DE Jr. Accuracy of angiographic quantification of peripheral atherosclerosis. Prog Cardiovasc Dis 1983;26:223– 236. Vogt KC, Just S, Rasmussen JG, Schroeder TV. Prediction of outcome after femoropopliteal balloon angioplasty by intravascular ultrasound. Eur J Vasc Endovasc Surg 1997;13:563–568.