Validation of the in vivo intravascular ultrasound measurement of in-stent neointimal hyperplasia volumes

Animal study protocol

This study was approved by the Animal Care and Use Committee of the Medlantic Research Institute and conformed to the tenets of the American Heart Association on research animal use. Sixteen specific-pathogen-free domestic pigs weighing 45 to 50 kg (Hidden Valley Farm, Reisterstown, MD) were pretreated with oral aspirin 325 mg, ticlopidine 250 mg and verapamil SR 120 mg 1 day before the procedure. On the day of the procedure, the animals were anesthetized with intramuscular ketamine (20 mg/kg, Fort Dodge Laboratories) and xylazine (2 mg/kg, Phoenix Pharmaceutical Inc., St. Joseph, MO) and intubated. Supplemental oxygen (2 liters/min) was administered continuously via a respirator, and sodium pentobarbital (10 mg/kg) was injected intravenously as needed. Under sterile conditions, the left carotid artery and external jugular vein were surgically exposed and 8Fr and 6Fr sheaths, respectively, were inserted. Continuous hemodynamic and surface electrocardiographic monitoring was maintained throughout the procedure (Marquette Electronics, Inc.). After systemic heparinization (300 U/kg intravenously), control angiograms of the left coronary arteries were performed with FL 3.5 guiding catheters (SciMed Life Systems, Inc.) using nonionic contrast agent (Optiray 320, Mallinckrodt Medical Inc., Maple Grove, MN) in two orthogonal views. A balloon catheter with one 15-mm Palmaz–Schatz stent (Cordis A. Johnson and Johnson Company, Warren, NJ) was placed in the proximal left anterior descending artery. The stent was deployed by inflating the balloon to nominal pressure (8 atm) for 1 min with a stent:artery of 1.3:1. Repeat angiograms were obtained immediately after stent implantation. All equipment was removed and the artery and vein were ligated. Four weeks after the procedure the animals underwent repeat angiograms in the same orthogonal views as well as IVUS imaging. After sacrifice, the coronary arteries were perfusion-fixed for histologic analysis.

IVUS imaging protocol

Studies were performed using a commercially available system (Cardiovascular Imaging Systems, Inc., San Jose, CA) which incorporated a single element beveled 30 MHz transducer mounted on the end of a flexible shaft and rotated at 1800 rpm within a 3.2Fr short monorail imaging sheath. Intravascular ultrasound imaging was performed after administration of 0.2 mg intracoronary nitroglycerin. The ultrasound catheter was advanced at least 10 mm beyond the stent and an imaging run (using automated transducer pullback at 0.5 mm/s) was performed to a point at least 10 mm proximal to the stent. Motorized transducer pullback through a stationary imaging sheath permitted the transducer to move at the same speed as the proximal end of the catheter; this has been validated in vivo (5). During imaging care was taken to set the overall gain and the time-gain-compensation curve to avoid suppressing the echolucent neointimal tissue present in the near field. Studies were recorded only during transducer pullback onto 1/2 inch high resolution s-VHS videotape for off-line analysis.

Quantitative coronary angiography (QCA) analysis

The QCA was performed by an independent core angiographic laboratory blinded to the IVUS and histomorphometric results. Quantitative coronary angiography was performed using an automated edge detection algorithm (CASS II, Pie Medical Imaging B.V., The Netherlands). Minimal lumen diameter (MLD), user-defined reference diameter and percent diameter stenosis (DS) after stent implantation and on follow-up were measured from multiple projections and the results from the “worst” view were recorded. Late lumen loss was calculated as postintervention minus follow-up MLD. The stented segment was then divided into seven sections. Minimum lumen cross-sectional areas (CSA) within each section were compared to the user-defined reference segment to calculate the distribution of neointimal tissue over the length of the stent.

Quantitative IVUS analysis

Intravascular ultrasound analysis was performed by an independent core laboratory blinded to the QCA and histomorphometric results. Motorized transducer pullback through a stationary imaging sheath permitted measurements of stent and lumen CSAs at 1 mm axial increments throughout the length of the stent. At a pullback speed of 0.5 mm/s, 1 mm is equivalent to 2 s of videotape. The stent CSA was measured at the leading edge of the stainless steel wires. If the neointimal tissue (IH) appeared to encompass the imaging catheter, the lumen CSA was assumed to be the physical (not acoustical) size of the imaging catheter (1.0 mm²). The IH CSA was then calculated as stent minus lumen CSA. Intimal hyperplasia volumes were then calculated using Simpson’s rule. These methods have been reported previously and are illustrated in Figure 1(3,4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1

The left-hand panelillustrates the location of the 15 IVUS image slices sampled within each Palmaz–Schatz stent. The middle panelshows a representative IVUS image slice. This slice is duplicated and annotated in the right-hand panel,indicating the stent CSA (outer white circle)and the lumen CSA (inner white circle). The IH CSA was calculated as stent minus lumen CSA.

Histomorphometric analysis

The perfusion-fixed hearts were harvested, maintained for 24 h in 10% buffered formalin and sent to the study pathologist. Then the specimens were embedded in methyl-methacrylate. Sections were cut with a low-speed diamond wafer mounted to a Buehler Isomet saw (Buehler Ltd., Evanston, IL), leaving the stent wires intact in the cross sections to minimize potential artifacts from removal of stent wires. Sections 50-μ to 100-μ thick were obtained at about 1 mm apart and stained with methylchromatic stain. This resulted in 15 to 20 cross sections per stent. Measurements were carried out on each section with the Sigmascan software (Jandel Scientific, San Rafael, CA) through an optical microscope integrated to a digitizing tablet. Histomorphometric analysis of each section included the neointimal, stent and lumen CSA. Using Simpson’s rule, neointimal tissue volume was calculated for each stent. These measurements were then compared with IVUS measurements.

Statistical analysis

Statistical analysis was performed using Statview 4.02 (Abacus Concepts) or SAS (Statistical Analysis Systems, SAS Institute Inc.). Continuous data were presented as mean value ± 1 SD. Intravascular ultrasound histomorphometric and QCA results were compared using linear regression analysis. The distribution of neointimal tissue over the length of the stent was determined by analysis of variance. A value p < 0.05 was considered statistically significant.

For each stent, five segments were identified and assigned to one of these categories: 1) distal edge, 2) distal body, 3) central articulation, 4) proximal body, 5) proximal edge. For each slice, the stent and lumen CSA was measured and neointimal tissue calculated (IH = stent-lumen CSA). The stent, lumen and IH CSA for these five slices were then compared using analysis of variance.

Sample size calculations were performed using the mean and standard deviation values from the entire study population to show a 25% and 50% reduction in IVUS IH volume, follow-up % diameter stenosis, late lumen loss, binary restenosis rate, alpha = 0.05 and 90% power (6).

Histomorphometric results

At follow-up, IH volume measured 26.7 ± 8.5 mm³by histomorphometry. Intimal hyperplasia was concentric and distributed uniformly over the length of the stent (Fig. 2). In 5 of 16 stents, the mean neointimal thickness was ≤0.2 mm.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

The distribution of IH CSA (millimeter squared) measured by IVUS (left-hand panel)and histopathology (right-hand panel)over the length of the stent is shown. Each stent is represented by 15 IVUS and histomorphometric measurements. The proximal end of the stent is on the left (beginning with slice #1 for IVUS and histomorphometry).

QCA results

By angiography, there was lumen narrowing throughout the stent. However, in all cases there was a focal region within the stented segment with smallest lumen, which was recorded as the minimum lumen diameter.

After stent implantation, reference vessel size was 2.66 ± 0.25 mm, MLD was 2.71 ± 0.29 mm and DS was −2.3 ± 11.5%. At follow-up, reference vessel size was 2.52 ± 0.30 mm, the mean MLD was 1.68 ± 0.53 mm, the mean lumen diameter was 2.34 ± 0.16 mm, the maximum lumen diameter was 2.63 ± 0.72 mm, and %DS was 30.3 ± 11.9%. Late lumen loss was measured as 1.03 ± 0.66 mm. Binary restenosis rate was 20%. Using the QCA measurement of follow-up %DS, sample size calculations showed that 15 stented lesions/arm would be required to show a 50% reduction in follow-up %DS; 52 stented lesions/arm would be required to show a 25% reduction in follow-up %DS. Using measurement of late lumen loss, 35 stented lesions/arm would be required to show a 50% reduction in late lumen loss; 139 stented lesions/arm would be required to show a 25% reduction in late lumen loss. Conversely, 266 stented lesions/arm would be required to show a 50% reduction in binary restenosis rate; 1,212 stented lesions/arm would be required to show a 25% reduction in binary restenosis rate.

IVUS results

Like the histomorphometric findings, IH was concentric and distributed uniformly over the length of the stent (Fig. 2). Intimal hyperplasia volume measured 30.4 ± 11.0 mm³by IVUS and correlated strongly with histomorphometry (r = 0.965, p < 0.0001, Fig. 3). The regression line was IVUS = 1.3 × histomorphometry −3.0. Intravascular ultrasound detected all five cases with a mean histomorphometric IH thickness ≤0.2 mm.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

The left-hand panelshows the correlation between the IVUS and histomorphometric measurements of IH volumes (IVUS = 1.3 × histomorphometry −3.0, r = 0.965, p < 0.0001). The right-hand panelshows the Bland and Altmann plot of IVUS-histomorphometric measurement of IH volumes. In all but one case, the IVUS measurement of IH volume was greater than histopathology.

The difference between the IVUS and the histomorphometric measurements of IVUS volume was 4.1 ± 2.7 mm³or 15.8 ± 11% (standard error of the estimate = 0.7). The Bland and Altman plot of this difference is shown in Figure 3. In all but one case, the IVUS measurement of IH volume was greater than histopathology. Table 1shows the mean and maximum lumen and stent diameters as measured by QCA, IVUS and histology within stented segments.

Using the IVUS measurement of IH volume, sample size calculations showed that 12 stented lesions/arm would be required to show a 50% reduction in IVUS-measured IH volume; 44 stented lesions/arm would be required to show a 25% reduction in IVUS-measured IH volume. Compared to QCA-based sample size calculations of follow-up %DS, late lumen loss and binary restenosis rate, sample size calculations based on IVUS measurements require a fewer number of stented lesions.

Technical considerations

Neointimal tissue is typically echolucent. Therefore, neointimal tissue may be more easily missed (or suppressed) than echo-dense structures such as stainless steel stent wires. To study the process of in-stent restenosis, it is important to optimize machine settings to avoid suppressing the nearfield echoes. Specifically, in the current study the overall gain and time-gain-compensation curves were set so that spontaneous blood echoes were visible in the lumen and so that the leading edge of the neointima was distinct.

Because metal is echoreflective, the stent served as both an ultrasonic and a histologic marker for the lesion. This facilitated the comparison of the IVUS images obtained in vivo, with the subsequent histomorphometric analysis performed ex vivo. Similar in vivo validation studies in nonstented lesions would be more difficult because of the lack of distinct ultrasound/histologic markers.

Limitations

Neointimal tissue accumulation and sample size calculations performed in animal preparations may be different compared to humans. However, the mean IH volume in the current study (30 mm³/Palmaz–Schatz stent) was similar to that reported in placebo groups in one human trial (7)as well as in a series of 26 consecutive lesions treated with single Palmaz–Schatz stents and imaged with serial IVUS (10). Similarly, sample size calculations based on one stent per lesion may be different when studying lesions treated with more than one stent. Multiple stents will result in more neointimal tissue per lesion (4). However, in the SCRIPPS Trial, there were an average of 1.5 stents/lesion; and despite the small number of patients, IVUS analysis showed a significant treatment effect (7).

Despite the excellent correlation (r = 0.965), IVUS measurement of IH volumes was larger than histomorphometry in all but one case. This tendency has been reported by others (11–13). In one previous study of directional coronary atherectomy in vitro, IVUS measurement of retrieved plaque volume was larger than that measured by histomorphometry but similar to tissue volume measured by water displacement (11). It is difficult to avoid completely distortion and shrinkage of the stented lesion during preparation for histomorphometric analysis. This could have accounted for the consistent difference between histologic and IVUS measurements of IH volume.

The current study measured image slices at 1-mm intervals. In this porcine coronary in-stent restenosis model, neointimal tissue was distributed uniformly over the length of the stent. However, in humans there may be a greater variation in the patterns of neointimal tissue accumulation (focal vs. diffuse, central vs. marginal focal, etc.). Thus, an analysis of 15 slices/15-mm long stent may undersample the variation in neointimal tissue in humans; and sample size calculations may also differ. An alternative approach would be the use of computerized automatic contour detection; such a system, which is already in clinical use, can sample 200 images/lesion (14–16). However, the current methodology (of 1-mm slices) has been used in one multicenter trial (7).

Sampling frequency may be dependent on vessel size. At or beyond the focal zone of the imaging catheters used in the current study, the beam may spread to 300 to 700 μ, limiting the ability to have independent samples.

Angiography cannot detect radiolucent stents. Therefore, the used-defined reference lumen CSA was used as a surrogate of stent area to calculate IH area and distribution.

Conclusion

Intravascular ultrasound measurements of neointimal hyperplasia volumes correlate well with histomorphometry. These measurements are valid for assessing strategies to reduce in-stent neointimal tissue accumulation and in-stent restenosis.