Comparison of Coronary Luminal Quantification Obtained From Intracoronary Ultrasound and Both Geometric and Videodensitometric Quantitative Angiography Before and After Balloon Angioplasty and Directional Atherectomy

Although QCA is the gold standard in interventional cardiology, pathology studies indicate that angiography may underestimate the extent and severity of atherosclerotic disease.¹² While ICUS provides unique information regarding vessel wall morphology compared with angiography,³⁴⁵⁶⁷⁸ precise quantitative analysis of luminal CSA by ICUS would offer a significant advantage in the guidance of coronary intervention procedures.⁹¹⁰¹¹ A recent multicenter ICUS study in patients with coronary angioplasty indicated that postinterventional luminal dimensions obtained by ICUS but not QCA may be a significant predictor of restenosis at follow-up.¹² Additionally, another multicenter study suggests that a large residual plaque burden remains on ICUS imaging, despite optimal angiographic results.¹³ Nakamura et al⁹ and Colombo et al¹⁰ also suggest that luminal measurements provided by ICUS may be helpful for optimal stent deployment. Previous studies, however, have provided conflicting evidence on the agreement between quantitative measurements derived from ICUS and those derived from QCA.^34714151617 The aim of our study was to clarify whether ICUS measurements agree with QCA measurements and to determine which factors, if any, may play a role in any discordance between the techniques. To do this, we compared MCSA obtained from ICUS and both ED and VID computer-based QCA before and after BA and DCA.

Methods

Patients

Patients who had an ICUS examination before and after single-vessel BA (n=100) or DCA (n=50) with adequate quality of ICUS and angiographic images for quantitative analysis were enrolled in this study. Preintervention, the US catheter did not cross the target lesions in 20 lesions because of proximal vessel tortuosity, the severity of the target stenosis, or transient serious arrhythmia during the ICUS examination. Additionally, the ultrasound catheter completely occluded the coronary lesion in 89 lesions. Thus, preintervention MCSA of the stenotic segment was measured in the remaining 41 lesions. Postintervention ICUS measurements were obtained in all 150 lesions in the 150 patients.

BA or DCA Procedures

All patients received full anticoagulant therapy including intravenous aspirin and heparin before ICUS examination and intervention. Coronary angiograms were recorded on cinefilm after the intracoronary administration of isosorbide dinitrate (1 to 2 mg). The size of the balloon or atherectomy device was determined to match the vessel RD obtained from the online QCA measurement. Luminal damage postintervention was qualitatively graded into three categories by angiographic assessment as none (smooth lumen), generalized haziness, or dissection. Dissection was defined according to the dissection classification types B, C, D, E, and F of the classification of the National Heart, Lung, and Blood Institute.¹⁸

ED-QCA

The new version of the computer-based Coronary Angiography Analysis System (CAAS II)¹⁹²⁰ was used to perform the ED and VID quantitative analyses. In the CAAS analysis,^192021222324 the entire 18×24-mm cineframe is digitized at a resolution of 1329×1772 pixels, and the boundaries of a selected coronary segment are detected automatically. The absolute diameters of the stenosis (MLD and RD) are determined by using the contrast-free guiding catheter as a scaling device. To standardize the method of analysis before and after intervention, all study frames selected for analysis were end-diastolic to minimize motion artifact, and arterial segments were measured between the same identifiable branch points in multiple views after the administration of isosorbide dinitrate.^2021222324 MCSA was calculated as π×(MLD1)×(MLD2)÷4 from measurements obtained from the ED analysis in orthogonal views (MLD1 and MLD2) before and after intervention.

VID-QCA

VID measurement is based on the relationship between the attenuating power of the lumen filled with contrast medium and the x-ray image intensity.²⁵ Using this relationship, a VID profile that was proportional to the CSA of the lumen was obtained. Subtraction of patient structure noise was applied after computing the linear regression line through the background pixels located on both sides of the detected luminal contours. Consecutive densitometric profiles of the analyzed segment were acquired in all scan lines perpendicular to the vessel including lesion, reference, and nondiseased areas. Conversion of the individual VID profiles to absolute values was performed after a transformation of the VID profile found in a CSA of a nondiseased segment, assuming a CSA at any point is proportional to the densitometric profiles at that point. MCSA was calculated from the average value obtained from the VID system in multiple views. The basic principles of the technique are illustrated in Fig 1.

ICUS Image Acquisition

Following angiography, an ICUS catheter (30 MHz; 2.9F, 3.2F, or 4.3F; Cardiovascular Imaging Systems) was introduced over a 0.014-in. guide wire and positioned distal to the lesion. Lesion geometry was then imaged by using a slow, continuous catheter pull-back procedure. Catheter position was documented by simultaneous fluoroscopy superimposed on the ICUS display screen. ICUS images were stored on super VHS tape for offline analysis.

Quantitative and Qualitative Assessment of ICUS

Luminal CSA was defined as the integrated area central to the intimal leading-edge echo. Images with MCSA were selected from the pull-back sequence by reviewing the position of the ICUS catheter on the angiographic image that was recorded on the same ICUS image and by reviewing the time log and audio recording of the procedure to analyze the same coronary segment as the quantitative angiogram. Total vessel CSA was defined as the area inside the interface between the plaque-media complex and adventitia (ie, the area inside the external elastic membrane). When the dissected lumen communicated constantly with the true lumen, the dissected lumen was included in the luminal area, as exemplified in Fig 2. Echo reflectivity was categorized as either low or high (plaque reflectivity lower or higher, respectively, than the bright adventitial layer).²⁶ Calcium deposits were defined as highly echo-reflective tissue with acoustic shadowing. A lesion was considered homogeneous if the plaque consisted of >75% of one type of echo reflectivity. A lesion was defined as mixed if it contained both high and low echo-reflective areas occupying >25% of the plaque area.²⁶ A lesion was considered predominantly calcific if calcium occupied >180° of the vessel circumference.²⁶

Luminal damage postintervention was qualitatively graded into three categories: regular lumen, irregular lumen including a small tear not extending to the media, and dissected lumen with circumferential tear behind the plaque or tear extending to the media.⁵⁶²⁷ The eccentricity ratio was calculated as the ratio between minimal and maximal wall thickness (1 indicates concentric plaque, <1 indicates increasing eccentricity).²⁷ To determine the interobserver variability of ICUS measurements, 30 videotapes of the complete original recording were used by two independent observers to select and measure the minimal CSA. The mean signed difference and correlation of the measurements of minimal CSA were −0.12±0.79 mm² and 0.94, respectively.

Statistical Analysis

In the absence of the known true values, Bland and Altman²⁸ recommend the use of the mean and SD of the signed differences between two measurement systems as an index of agreement between the two systems. Thus, we took the mean and SD of the signed differences between ICUS and QCA measurements as an index of agreement between ICUS and QCA measurements instead of linear regression analysis. The individual measurements obtained from ICUS and QCA were compared by using the paired Student t test and correlation coefficient. A probability value of <.05 was considered significant.

Results

Baseline Clinical and Angiographic Characteristics

No difference was found in gender, age, anginal symptoms, or distribution of diseased vessels between the BA and DCA groups (Table 1). QCA measurements were obtained in 100 lesions before and after BA and in 50 lesions before and after DCA. ED-QCA indicated that the RD before and after intervention and MLD postintervention were significantly larger in the DCA than the BA patients. MCSA was measured by using ICUS in 26 lesions before BA and 15 lesions before DCA without wedge of the ICUS catheter. One hundred lesions after BA and 50 lesions after DCA were estimated by using ICUS, which revealed that 64 lesions in BA and 31 lesions in DCA consisted of homogeneous plaque; all remaining lesions were classified as mixed. Most of the homogeneous plaque was low echo reflective. Focal calcium deposits were observed in 42 lesions in BA and 19 lesions in DCA, while moderate-to-diffuse calcification was seen in 32 lesions in BA and 16 lesions in DCA (P=NS). Total vessel CSA, plaque and medial CSA, and percent plaque and medial CSA preintervention were larger in the DCA than the BA group. Lesion eccentricity was not significantly different between the two groups.

MCSA Measured by ICUS, ED-QCA, and VID-QCA Before and After Interventions

Table 2 shows the MCSA measured by ICUS, ED-QCA, and VID-QCA before and after BA and DCA. MCSA in nonwedged lesions (BA=26 lesions and DCA=15 lesions) as obtained by ICUS was significantly larger than MCSA measured by using ED- or VID-QCA pre-BA (both P<.01) and pre-DCA (both P<.01). Post-BA MCSA in 100 lesions as obtained by ICUS was significantly larger than the MCSA measured by either ED- or VID-QCA (both P<.01). Post-DCA MCSA in 50 lesions as obtained by ICUS was significantly larger than MCSA as measured by ED-QCA (P<.01) but not VID-QCA.

Agreement Between ICUS, ED, and VID Before and After BA and DCA

Table 3 compares the agreement between the three measurement techniques, and Figs 3 and 4 display the postintervention agreement between measurements obtained from ICUS and ED-QCA (Fig 3) and from ICUS and VID-QCA (Fig 4). The correlation coefficient between the ICUS and ED measurements decreased from .59 pre-BA to .47 post-BA and from .57 pre-DCA to .44 post-DCA. The absolute difference between ICUS and ED was significantly greater post-BA and post-DCA than pre-BA and pre-DCA (both P<.05). The agreement between ICUS and ED deteriorated post-BA and DCA compared with the pre-BA and DCA agreement. The correlation coefficient between ICUS and VID measurements increased from .50 pre-BA to .63 post-BA and from .50 pre-DCA to .72 post-DCA (Table 3). The difference between ICUS and VID was not significantly different from pre-BA and DCA to post-BA and DCA. Postintervention, the difference between ICUS and VID was significantly less than the difference between ICUS and ED post-BA and post-DCA (both P<.01). The discordance between ICUS and VID was smaller than the discordance between ICUS and ED both post-BA and post-DCA. While in both pre-BA and DCA no significant difference was observed between ED and VID, in post-BA and DCA there was a significant difference between the two (BA, P<.001; DCA, P<.01).

Luminal Damage Postintervention as Assessed by Angiography and ICUS

The degree of luminal damage postintervention as assessed by using angiography and ICUS is given in Table 4. Concordance between the two qualitative imaging techniques was found in 23 (70%) of the 33 patients with angiographically detected dissected lesions and 23 (66%) of the 35 patients with dissection as detected by ICUS post-BA and in 10 (83%) of the 12 patients with angiographically detected dissected lesions and 10 (59%) of the 17 patients with dissection as detected by ICUS post-DCA.

Influence of Vessel Damage Induced by BA and DCA in the Agreement Between ICUS and QCA

The correlation coefficient of ICUS and ED quantitative measurements was .74 in lesions with an angiographically determined smooth lumen, .46 in lesions with angiographic haziness, and .26 in lesions with angiographic evidence of dissection (Table 5). The correlation of ICUS and ED quantitative measurements was .70 in lesions with a regular lumen as determined by ICUS, .52 in lesions with an irregular lumen, and .10 in lesions with ICUS evidence of dissection. Thus, the presence of vessel damage induced by BA was associated with a deterioration of agreement between ICUS and ED measurements. While agreement between ICUS and VID-QCA also deteriorated in the presence of morphological changes induced by BA, the difference of the measurements between ICUS and VID was significantly less than the difference between ICUS and ED in the presence of both angiographic (P<.05) and ICUS (P<.05) evidence of dissection. While high agreement was obtained between ED-QCA and VID-QCA, agreement between these two techniques also decreased post-BA according to the increase of vessel damage.

A similar pattern was seen when lesions treated by DCA were categorized according to their morphological characteristics (Table 6). Poor agreement was obtained in lesions with angiographic or ICUS evidence of vessel damage compared with lesions with an angiographically smooth lumen or ICUS appearance of a regular lumen. The absolute difference of the measurement between ICUS and VID was significantly less than the difference between ICUS and ED in the presence of both angiographic (P<.01) and ICUS (P<.01) evidence of dissection.

Correlation Between QCA Analyses of the Same Lesion From Multiple Views

To ensure that the better relationship between VID and ICUS was a true phenomenon and not due to a greater variation in values obtained from different views, we looked at the correlation and differences between orthogonal measurements for both VID and ED before and after intervention.²⁹³⁰³¹ The correlation and differences between orthogonal measurements obtained by ED were 0.69 (0.21±0.62 mm²) preintervention and 0.49 (0.39±2.33 mm²) postintervention. The values obtained for VID were 0.69 (0.14±0.71 mm²) preintervention and 0.67 (−0.33±1.79 mm²) postintervention.

Discussion

The principle findings of our study were (1) that MCSA obtained by ICUS was significantly larger than MCSA as measured by either ED- or VID-QCA both before and after BA and DCA, (2) that the agreement between ICUS and ED deteriorated considerably after both BA and DCA, (3) that the complex morphological changes induced by BA and DCA contributed to the discordance of the agreement between ICUS and ED, and (4) that VID measurements were found to provide a better agreement with ICUS than ED measurements, particularly in lesions with complex morphological changes post-BA and DCA.

Agreement Between ICUS and ED-QCA in Previous Studies

Previous studies have provided conflicting evidence on whether luminal measurements obtained from ICUS agree with ED-QCA measurements in human coronary arteries. In general, studies that examined ICUS and ED measurements in normal coronary segments report a favorable correlation between the two quantitative imaging modalities,³⁷ while those that examined lesions postangioplasty report a poor correlation.⁴¹⁵¹⁶ Nakamura and colleagues¹⁷ report that dissection induced by BA plays a role in the discordance between ICUS and ED-QCA measurements, but there are several methodological differences between their study and ours. First, they did not examine VID measurements. Second, they showed only the correlation coefficient in groups with and without dissection, and no statistical difference was found in the difference of the individual ICUS and angiographic measurements between two groups. Finally, they did not use a computer-based QCA system but rather caliper measurements, which have a poor reproducibility and result in frequent underestimation or overestimation of stenosis severity.³² Our study is the first to compare both ED- and VID-QCA measurements with ICUS measurements.

Factors Contributing to the Discordance of ICUS and QCA Measurements

Agreement of luminal area measurements as obtained by using ICUS and ED deteriorated considerably after BA and DCA. A progressive deterioration in the relationship between the ICUS and QCA measurements was seen in accordance with the presence of increasing vessel damage and increasing luminal complexity postintervention. Thus, the cross-sectional shape of the vessel lumen postintervention may be a significant factor in the discrepancy between ICUS and ED-QCA measurements. MLD obtained from ED-QCA depends on the angiographic projection. Although we calculated luminal area from two orthogonal views, the chances of obtaining the exact minimal and maximal diameters of the lesion CSA using ED-QCA would be small, particularly in the complex elliptical shape of the lumen postintervention (Fig 5). ICUS and VID are not projection dependent, and both would provide a measure of the “depth” as well as the “width” of the lumen cross section. ED, however, provides only a measure of one diameter (the “width”) of the lumen. Consistent with this is the fact that VID-QCA provided a better agreement and less relative underestimation in relation to ICUS measurements than ED-QCA.

The underestimation by ED relative to ICUS measurements may reflect the propensity of the contour-detection algorithm to trace the change in the brightness profile in the contrast-weak channel between the true and false lumens³³ (Fig 2), while ICUS measurements and VID measurements in multiple views may include the contribution of the false lumen. Such a phenomenon, however, would not account for the relative underestimation by ED in the preintervention phase. Another possibility is that the interventional cardiologist tends to select angiographic projections that best demonstrate both the stenosis preintervention and the residual stenosis postintervention, ie, “worst view” angiography. Even using multiple orthogonal views, as we did in our study, the assumed elliptical cross section may have been based on multiple “worst views,” which although they were, per protocol, >45° apart, were not necessarily a combination of truly “worst view” and “best view.” Such a limitation to ED-QCA has been an inherent problem in all coronary interventional trials, and attempts to address this by three-dimensional imaging in truly orthogonal views are currently under evaluation.³⁴³⁵

Two additional ICUS-related factors may also have contributed to the observed discordance between ICUS and quantitative angiographic measurements. Elliptical angulation of the ultrasound catheter within the longitudinal axis of the vessel may have led to overestimation of luminal dimensions by ICUS. Additionally, introduction of the ultrasound catheter may have itself resulted in tacking back of dissection flaps, with a resultant larger lumen during ICUS examinations postintervention compared with the less invasive technique of contrast angiography. Our data also suggest that quantitative angiography may yield larger luminal measurements than ICUS in a significant number of patients (Figs 3 and 4). This increase in the apparent angiographic diameter may be caused by extraluminal contrast within fissures, cracks, and dissection as seen around the true lumen.

Study Limitations

First, the coronary sites compared by ICUS and QCA may not have been exactly identical. Although we tried to ensure that this was the case by using simultaneous recording of fluoroscopy and ICUS imaging as well as landmarks such as side branches to guide us, there is no guarantee that we analyzed exactly the same point of the coronary artery in ICUS and QCA measurements. This is, however, a generic problem of any ICUS-QCA study³⁴ that would be very difficult to overcome, as the presence of the ICUS catheter at the lesion site during coronary arteriography would interfere with the QCA measurements. Second, both ED- and VID-QCA analysis were performed using only the CAAS II system. Thus, further studies would be required to confirm if our findings can be generalized to other QCA hardware or software systems.²¹ It is conceivable that if an ED algorithm were inaccurate in the normal reference segment, then a systematic underestimation or overestimation of vessel diameters could be translated to subsequent VID measurements. Third, it is also possible that ultrasound image analysis fails to see the true leading intimal edge, especially if the plaque has a low fibrous component and appears relatively hypoechoic, thus overestimating luminal dimensions. Additionally, a poor dynamic range can also induce technical intimal drop-out, leading to lumen overestimation.

Clinical Implications

Although ICUS provides unique information regarding the vessel wall morphology, the clinical utility of this technique remains unproven to date. ICUS luminal measurements, however, do provide important additional information to that obtained by visual assessment for optimal stent implantation.⁹¹⁰ Whether ICUS provides more accurate information than QCA, however, has not yet been determined. Additionally, whether luminal measurements obtained from ICUS are a superior index for the short- and long-term success of interventional procedures awaits the results of recent multicenter trials.¹²¹³³⁶ While ICUS is not universally available and involves additional time and expense,¹¹ QCA is more widely available and less time-consuming. Our data suggest that in the absence of the known true value and assuming that ICUS gives the most accurate estimate of luminal dimensions, QCA measurements, particularly ED, may be compromised, especially in assessing the complex luminal morphology following BA or DCA. Our study suggests that VID may offer a better correlation with the true luminal dimensions, as reflected by ICUS, than ED-QCA. VID may thus be the “poor man’s” ICUS, especially in lesions with complex morphology.

Conclusions

MCSA measurements obtained by ICUS are significantly larger than measurements provided by either geometric or VID-QCA both before and after intervention. Agreement between ICUS and geometric QCA measurements deteriorate considerably after intervention. Complex morphological changes induced by intervention may play a role in such a discordance between the two quantitative imaging techniques. In the absence of ICUS, VID, which is currently available in an online QCA system, may provide a better alternative than ED-QCA in lesions with complex morphology.

Selected Abbreviations and Acronyms

Presented in part at the 43rd Annual Scientific Session of the American College of Cardiology, Atlanta, Ga, March 13-17, 1994, and previously published in abstract form (J Am Coll Cardiol. 1994;23[suppl]:70A).

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

• Download figure
• Download PowerPoint

Footnotes