Lesion Characteristics and Coronary Stent Selection with Computed Tomographic Coronary Angiography: A Pilot Investigation Compar

Lesion Characteristics and Coronary Stent Selection with Computed Tomographic Coronary Angiography: A Pilot Investigation Comparing CTA, QCA and IVUS

   ABSTRACT: Objective. The accurate assessment of a target coronary lesion and appropriate stent selection is important in ensuring procedural success during percutaneous coronary intervention (PCI). Though quantitative coronary angiography (QCA) and intravascular ultrasound (IVUS) are available, stent selection is most commonly performed by visual estimation alone. Computed tomographic coronary angiography (CTA) has been shown to correlate well with QCA and IVUS in the assessment of coronary stenoses and may also have a role in stent guidance. Materials and Methods. Patients awaiting elective PCI underwent CTA. Blinded observers assessed lesion characteristics using: CTA, QCA, IVUS and visual estimation. Luminal diameters, lesion lengths, ACC/AHA lesion types and CTA-suggested stent sizes were compared. Results. A total of 17 patients (26 lesions) were evaluated. There was good correlation between CTA and IVUS for luminal diameter and for lesion length (r = 0.86 and 0.71, respectively). Similarly, the inter-test variability between the two methods using the intra-class coefficient (ICC = 0.85) was similar to the inter-observer variability of IVUS (ICC = 0.90). The agreement between CTA and visual estimation for lesion type was good (K = 0.79) and was similar to the agreement between the two visual observers (K = 0.72). There was good correlation between CTA stent recommended and actual stent selected (diameter, r = 0.82; length, r = 0.64). Conclusions. If CTA data is available prior to coronary angioplasty, the reporting of luminal size, length, and lesion type may assist the clinician with coronary stent selection.

J INVASIVE CARDIOL 2010;22:328–334

Key words: percutaneous coronary intervention; stent prescription; CT coronary angiography; intravascular ultrasound    Percutaneous coronary intervention (PCI) is the most common revascularization strategy for patients with symptomatic obstructive coronary artery disease (CAD). Procedural success is dependent on the accurate assessment of the target lesion and accurate stent selection. Though quantitative coronary angiography (QCA) and intravascular ultrasound (IVUS) are clinically available, stent selection is most commonly performed by visually estimating lesion length and luminal diameters at the time of angioplasty. Although much of the data are not recent, it is still well accepted that errors in visual estimation (VE) may lead to the oversizing or undersizing of stents and may result in a greater risk of in-stent restenosis, acute stent thrombosis or vessel rupture.^1–6 In addition to reducing procedural complications and improving outcomes,^7–11 accurate estimation of lesion length may reduce the total number of stents needed, in turn, minimizing procedural costs.    The simplicity, ease and convenience of VE are extremely attractive to the operator and although IVUS is available, cost and time prohibit its routine use in all patients. IVUS is more commonly reserved for cases where there is uncertainty in lesion dimensions or characteristics. Computed tomographic coronary angiography (CTA) is an emerging noninvasive imaging modality with good sensitivity and specificity for detecting obstructive CAD and plaque characteristics.¹² Given its ability to assess coronary anatomy as well as nearby coronary plaque, it may have the ability to provide additional information to assist with stent selection. In patients who have had initial CTA prior to PCI, additional information may be derived from the CTA dataset, which may better enhance stent size selection. The objective of this pilot study is to compare lesion characteristics that are required for better stent selection, as assessed with CTA, VE, QCA and IVUS. We also prospectively compared the stent recommended using CTA to the actual stent deployed. Materials and Methods    Stable patients undergoing elective coronary angioplasty and stenting at the study institution were prospectively identified and underwent CTA prior to PCI.    Patients with acute coronary syndrome were excluded, as were patients with renal insufficiency (creatinine > 135 µmol/L [> 1.5 mg/dl]), atrial fibrillation, a history of coronary revascularization, the inability to perform a breath-hold or contraindications to contrast or radiation exposure. All patients provided informed consent and the study protocol was approved by the institutional Human Research Ethics Board. Computed Tomographic Coronary Angiography: Image Acquisition Protocol    Prior to image acquisition, a heart rate of 65 beats per minute was targeted with the use of beta-blockers or calcium channel-blockers. Sublingual nitroglycerin (0.8 mg) was administered prior to image acquisition.    A biphasic timing bolus (15–25 cc contrast and 40 cc saline) was used to optimize the timing between intravenous contrast infusion (iodixanol 320 or iohexol 350) and image acquisition. Final images were acquired with a triphasic protocol (100% contrast, 40%/60% contrast/saline, and 40 cc saline). The contrast volume and infusion rate (5–8 cc/sec) were individualized according to scan time and patient body habitus.    Retrospective ECG-gated datasets were acquired with the GE Lightspeed Volume CT (GE Healthcare, Milwaukee, Wisconsin) with 64 x 0.625 mm slice collimation and a gantry rotation of 350 msec (mA = 300–800, kVp = 120). Scanning pitch (0.16–0.24) was individualized to the patient’s heart rate. Using the cardiac phase(s) with the least cardiac motion, CT datasets were reconstructed with a slice thickness of 0.625 mm and an increment of 0.4 mm. Computed Tomographic Coronary Angiogram Image Analysis    Images were processed offline using the GE Advantage Volume Share Workstation (GE Healthcare) and interpreted by two expert observers blinded to all clinical data prior to PCI. All final measurements were based on consensus between the two readers. Using oblique multiplanar reformations, lesion characteristics were assessed manually in both short and long axes using optimized window width and level. Similar to the recommendations published on IVUS measurements, the largest and “most normal” luminal diameters proximal (PLD) and distal (DLD) to the stenosis, as well as the lesion lengths, were measured (Figure 1).¹³ Measurements were performed in the portion of the artery with the least amount of artifact (e.g., beam hardening).    Based on the measured dimensions and surrounding calcific and noncalcific plaque, an ACC/AHA lesion type (based on lesion length, angulation, calcification, etc.) was assigned and a stent size was suggested.¹⁴ Final stent selection was left to the discretion of the interventionist who had access to the procedural IVUS results. All decisions were made blinded to the CTA results. Invasive Coronary Angiogram Analysis (Visual Estimation, Quantitative Coronary Angiography and Intravascular Ultrasound)    Prior to PCI, multiple oblique views of the target lesion were acquired with the Elite Allura Xper FD10 G-arm (Philips, Netherlands). Intracoronary nitroglycerin (100–400 mcg) was administered and IVUS was performed using the Atlantis SR 3F 40 MHz (Boston Scientific Corp., Natick, Masachusetts) with an automated motorized pullback (0.5 mm/sec) system. Images were analyzed using the Boston Scientific Galaxy software. Based upon the IVUS measurements, a stent was selected by the angioplasty operator and was deployed.    For the purpose of data analysis, all images acquired at the time of PCI were re-analyzed offline by expert observers blinded to all clinical data and previous measurements. Angiographic images were visually assessed by two expert readers (VE_obs1 and VE_obs2) blinded to the CTA images for luminal diameters, lesion lengths and ACC/AHA lesion types.¹⁴ IVUS data were analyzed by two expert observers (IVUS_obs1 and IVUS_obs2) blinded to the CTA data based on published criteria,¹³ and discrepancies were resolved by consensus. In short, length was calculated from short-axis images based on a 0.5 mm/sec pullback, whereas luminal dimensions were measured exclusively in the short axis. QCA was performed on the images used for coronary angioplasty by an expert observer blinded to the results of VE and IVUS.¹⁵    Statistical analysis. Continuous variables are presented as means with standard deviations or medians with interquartile ranges. Categorical variables are presented as frequencies with percentages. Correlations of continuous variables were compared using the Pearson product moment correlation coefficient. For correlation analysis, the mean VE (VEmean) and consensus IVUS measurements were used. Analysis was performed in all interpretable lesions. A two-tailed paired-samples student’s t-test was utilized to assess for statistically significant differences within continuous variables.    To better understand the potential significance of the discrepancies observed between CTA and IVUS, the intertest variability using the intraclass correlation coefficient (a method of agreement for continuous variables) between CTA and IVUS was compared to the interobserver variability for IVUS (IVUS_obs1 and IVUS_obs2) and for visual estimation (VE_obs1 and VE_obs2). Differences in the intraclass correlation coefficient (ICC) were considered statistically significant if there was no overlap of the 95% confidence intervals.    Agreements in measurements were plotted using the method described by Bland and Altman.¹⁵ Kappa agreements were performed for noncontinuous variables. Differences for continuous variables were considered statistically significant when p Results    A total of 18 patients were prospectively recruited to the study. One patient was excluded because of inadequate intravenous access. The remaining 17 patients (mean age = 63 ± 11 years; 59% men) underwent both CTA and PCI (Table 1). CTA was performed a median of 5 days before angioplasty (IQR: 2, 36). Three patients had suboptimal CTA images due to misregistration artifact or motion artifact, but all 3 datasets were still ultimately interpretable.    All lesions were successfully imaged by IVUS, however 3 patients with critical stenoses required minimal balloon predilatation prior to the advancement of the IVUS catheter. Though balloon predilatation may affect plaque size and distribution, we measured luminal dimensions in the “most normal” proximal and distal segments¹² and after nitroglycerin-induced vasodilatation. Since the potential effect of predilatation on a maximally dilated normal lumen is likely minimal, these patients were included in the study analysis. Correlation of lesion lumen, length and type. The predominant coronary plaque encountered was fibrous, although 4 lesions had significant calcification. The median luminal diameter according to IVUS was 3.1 mm (IQR: 2.5, 3.6), with a length of 9.9 mm (IQR: 5.9, 13.9) (Table 2). The median plaque attenuation (Hounsfield units) by CTA was 81 (IQR: 61, 126).    There was very good correlation between CTA and IVUS for all luminal diameters (r = 0.86), PLD and DLD (r = 0.80 and r = 0.91) (Figure 2A). There was no statistical difference between CTA and IVUS for luminal diameter (p = 0.262). Bland-Altman plots demonstrated no systematic bias between CTA and IVUS measurements of luminal diameter (Figure 2A). Similarly, there was good correlation for lesion length (r = 0.71), although lesion lengths were longer with CTA (p = 0.003) (Figure 2B). When the 4 heavily calcified lesions were excluded, the correlation improved slightly in both luminal diameter (r = 0.87) and length (r = 0.74). Figure 3 demonstrates a lesion in the left anterior descending artery with similar luminal measurements and a longer length measurement with CTA.    A similar analysis was performed comparing VE to IVUS and QCA to IVUS. Visual estimation (VE_obs1, VE_obs2, VEmean) correlated well with IVUS with respect to luminal diameter (0.72, 0.78, 0.79) and length (0.66, 0.76, 0.81) (Figures 4A and 4B). Similar to CTA, VE systematically overestimated the lesion length when compared to IVUS (p = 0.002). Luminal diameter as observed by visual estimation (2.98 ± 0.47 mm) was also smaller than that measured with IVUS (3.15 ± 0.82 mm; p = 0.019).    Conversely, QCA had fair correlation with IVUS for both luminal diameter (r = 0.62) and lesion length (r = 0.49). When compared to IVUS, luminal diameter was systematically smaller with QCA (2.39 ± 0.64 mm; p obs1 and IVUS_obs2 and between VE_obs1 and VE_obs2 for both luminal diameters and lesion lengths (Table 3). Furthermore, the ICC for luminal diameters between IVUS and CTA (0.85; CI: 0.76–0.91) appeared to be superior to the ICC between IVUS and VE (0.56; CI: 0.35–0.72) and the ICC between IVUS and QCA (0.26; CI: -0.01–0.50) (Table 2). There was no difference in the intertest variability with lesion lengths when comparing CTA, VE and QCA.    Assistance in stent selection. The correlation between CTA stent (diameter and length) recommendation and the final stent deployed (as guided by IVUS) was good (r = 0.82 and r = 0.64, respectively). The average stent dimension recommended by CTA was 3.0 x 15.7 mm, which is the same as the average stent initially selected, i.e., 3.0 x 15.7 mm. CTA prescribed a total of 26 stents (for 26 lesions) and a total length of 407 mm. The total number of stents used was 31, with a total stent length of 454 mm (p = 0.076 for difference). The extra stents were to cover an area of plaque which was not covered with the initial stent, or otherwise to deal with a complication. Discussion    PCI and stent selection are unique to each lesion being considered, taking into account the true vessel size of the nearby normal reference segments, the likely lesion length, as well as plaque composition. The choice of stent diameter can be variable and somewhat forgiving, as smaller stents can, within reason, be dilated to a larger size. However, undersizing and oversizing stent diameters may lead to in-stent restenosis or vessel rupture, respectively. Also, undersizing stent length may lead to coronary dissection as well as residual exposed plaque, both of which subsequently require further stenting accompanied by the inherent increased risks associated with multiple stent implantations.    Currently, IVUS is the gold standard for lesion characterization prior to coronary intervention. Though safe, several factors (cost, time and the lack of evidence that IVUS improves procedural success or patient outcomes) limit the routine use of IVUS in every patient.¹⁷ Though visual assessment still has a good correlation with IVUS, in complex anatomy, further imaging may be of benefit. The results of our study support the need for further investigation to better understand the potential incremental value of CTA prior to PCI.    A noninvasive, inexpensive and accurate method to assist with lesion characterization and subsequent stent selection is desirable. Though further studies are required, our pilot results suggest that CTA may provide similar measurements to those of IVUS for luminal diameter (p = 0.262). More importantly would be the undersizing of luminal diameter with both visual estimation as well as QCA when compared to IVUS which is a well identified risk factor for stent restenosis5 and thrombosis.⁶    The results of this study demonstrate that CTA may offer valuable information as an alternative to IVUS in patients with preexisting CTA data. This notion is further supported by the fact that the variability between CTA and IVUS was not different than the variability between IVUS_obs1 and IVUS_obs2 or VE_obs1 and VE_obs2.    The assessment of coronary stenoses with CTA has traditionally been performed visually since the current spatial resolution and partial volume effects are felt to preclude accurate assessment of a small lumen. More recent studies have quantitatively compared CTA to IVUS in the assessment of coronary stenoses and demonstrated that the mean coronary stenosis between both modalities were comparable (41.1% v. 50.4%; r = 0.61).¹⁸ One study examined 54 coronary segments with 30–70% stenoses and demonstrated that the correlation between CTA and IVUS for the measurement of minimal luminal cross-sectional area was excellent (r = 0.88).¹⁹ Similar to our results, the authors also found a poor correlation between QCA and IVUS when assessing lesion characteristics.    Given the superior spatial and temporal resolution, as well as the cross-sectional image display of IVUS, the luminal diameter measurements by IVUS remain the current gold standard. Given CTA’s current isotropic spatial resolution (0.40–0.50 mm), the measurement of a large “normal” reference lumen is likely accurate and accounts for the very good correlation between IVUS and CTA. The current spatial resolution, being a limitation of CTA, also explains the very good, but imperfect, correlation between IVUS and CTA.    Of concern were the discrepancies in lesion lengths. Compared to IVUS (mean of 10.9 mm), lesion lengths measured longer with both CTA (13.8 mm; p = 0.003) and VE (13.5 mm; p = 0.002). Such discrepancies may be partially explained by the subjective evaluation of plaque burden with the presence of adjacent plaque perceived as part of the culprit lesion. In addition, the measurement of lesion length with IVUS may be hampered by the longitudinal motion of the coronary tree throughout the cardiac cycle. This translational motion has been well observed as side branches move in and out of the IVUS imaging plane. Further limiting the accuracy of IVUS length measurements are the reliance on a “pullback” and the inability of IVUS to display true longitudinal images of the vessel (Figure 3). Conversely, CTA acquires 3-dimensional datasets with isotropic resolution and permits the visualization of the entire coronary vessel in all planes, allowing for cross-sectional and longitudinal measurements (Figure 3).    Although the preliminary results of this study are encouraging, the widespread use of CTA for stent selection is currently not practical nor recommended because of the excess radiation associated with this modality,²⁰ although with prospective imaging²¹ or reduced kVp,²² the radiation dose is reduced significantly without a notable compromise in interpretive quality.    However, in patients undergoing PCI who have preexisting CTA datasets, lesion assessment by the angioplasty operator may be possible for more accurate stent selection. At this time, CTA is best used in combination with VE. Agreement between VE and CTA would be reassuring to the operator, but significant discordance between the two modalities may identify lesions which may benefit from further assessment with IVUS.    Study limitations. The results of this small pilot study require confirmation. Aside from the small sample size, several obstacles (temporal resolution, spatial resolution, radiation, beam hardening and blooming artifacts) need to be addressed before CTA-guided stent selection can be routinely advocated. The authors acknowledge that the temporal and spatial resolution of CTA is inferior to that of invasive coronary angiography. Equally restrictive is the need for CTA readers to have an understanding of PCI and variables considered by interventional cardiologists before stent selection. Conversely, interventional cardiologists will require a better understanding of CT characteristics and potential shortcomings to make informed decisions. These factors may limit the general use of CTA for stent recommendation to experienced centers.    The number of lesions with heavy calcification was low in our study. It is likely that dense calcification would limit vessel characterization and subsequent accurate luminal measurements. This requires confirmation in a larger study.    Variations in luminal diameter measurements may have been influenced by the route of nitroglycerin administration with CTA (sublingual) and invasive coronary angiography (intracoronary), although no systematic bias was demonstrated in our study. Plasma levels have been shown to be higher with intravenous nitroglycerin than when administered orally or sublingually.²³ Furthermore, intracoronary nitroglycerin, in an angina population, has been shown to induce a greater amount of coronary vasodilatation when compared to the sublingual form.²⁴ The impact on the luminal dimensions measured with IVUS and CTA is uncertain and may account for some discrepancies between the modalities. Conclusions    In this small pilot study, CTA appears to be reasonably accurate in the assessment of coronary lesion dimensions. Though these results need to be confirmed in a larger study with more heterogeneous and complex lesions, we found that the correlation between CTA and IVUS was excellent for luminal diameter. Also, the intertest variability between CTA and IVUS was similar to the interobserver variability within VE and within IVUS. Although we cannot recommend that CTA be performed routinely prior to PCI, in cases where there is preexisting CTA data, it may be reasonable to review lesion characteristics which may assist the interventionist with stent selection. In cases where discrepancies between CTA and VE exist, further assessment with IVUS may be warranted.    Funding. Boston Scientific provided the IVUS catheters but had no input in the design of the study or the decision to publish. From the *Department of Medicine (Cardiology), University of Ottawa Heart Institute, Canada; the §University of Ottawa Heart Institute Cardiovascu- lar Methods Centre, Canada; and the ∞Department of Radiology, University of Ottawa, Canada. Disclosures: Benjamin Chow is supported by CIHR New Investigator Award #MSH-83718. Dr. Chow receives research support from GE Healthcare, Pfizer and AstraZeneca, fellowship training support from GE Healthcare and educa- tional support from TeraRecon Inc. This study was supported in part by Boston Scientific Corporation. Manuscript submitted February 11, 2010, provisional acceptance given March 22, 2010, final version accepted April 1, 2010. Address for correspondence: Dr. Benjamin J. W. Chow, University of Ottawa Heart Institute, 40 Ruskin Street, Ottawa, Ontario, Canada K1Y 4W7. E-mail: bchow@ottawaheart.ca References

1. Kuntz RE, Gibson CM, Nobuyoshi M, Baim DS. Generalized model of restenosis after conventional balloon angioplasty, stenting and directional atherectomy. J Am Coll Cardiol 1993;21:15–25. 2. Smith SC Jr, Feldman TE, Hirshfeld JW Jr, et al. ACC/AHA/SCAI 2005 Guideline Update for Percutaneous Coronary Intervention-Summary Article: A report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol 2006;47:216–235. 3. Mak KH, Belli G, Ellis SG, Moliterno DJ. Subacute stent thrombosis: Evolving issues and current concepts. J Am Coll Cardiol 1996;27:494–503. 4. Moussa I, Di Mario C, Reimers B, et al. Subacute stent thrombosis in the era of intravascular ultrasound-guided coronary stenting without anticoagulation: Frequency, predictors and clinical outcome. J Am Coll Cardiol 1997;29:6–12. 5. Kastrati A, Schömig A, Elezi S, et al. Predictive factors of restenosis after coronary stent placement. J Am Coll Cardiol 1997;30:1428–1436. 6. van Werkum JW, Heestermans AA, Zomer AC, et al. Predictors of coronary stent throm- bosis: The Dutch Stent Thrombosis Registry. J Am Coll Cardiol 2009;53:1399–1409. 7. Stone GW, Grines CL, Cox DA, et al. Comparison of angioplasty with stenting, with or without abciximab, in acute myocardial infarction. N Engl J Med 2002;346:957–966. 8. Muhlestein JB, Anderson JL, Cui C, et al. Improved long-term survival associated with stent deployment during percutaneous coronary interventions: Results from a registry of 3399 patients. Am Heart J 2005;150:182–187. 9. Serruys PW, de Jaegere P, Kiemeneij F, et al. A comparison of balloon-expandable- stent implantation with balloon angioplasty in patients with coronary artery disease. BENESTENT Study Group. N Engl J Med 1994;331:489–495. 10. Fischman DL, Leon MB, Baim DS, et al. A randomized comparison of coronary- stent placement and balloon angioplasty in the treatment of coronary artery disease. Stent Restenosis Study Investigators. N Engl J Med 1994;331:496–501. 11. Versaci F, Gaspardone A, Tomai F, et al. A comparison of coronary-artery stenting with angioplasty for isolated stenosis of the proximal left anterior descending coronary artery. N Engl J Med 1997;336:817–822. 12. Raff GL, Goldstein JA. Coronary angiography by computed tomography. J Am Coll Cardiol 2007;49:1830–1833. 13. Mintz GS, Nissen SE, Anderson WD, et al. ACC clinical expert consensus document on standards for the acquisition, measurement and reporting of intravascular ultra- sound studies: A report of the American College of Cardiology Task Force on Clinical Expert Consensus Documents (Committee to Develop a Clinical Expert Consensus Document on Standards for Acquisition, Measurement and Reporting of Intravas- cular Ultrasound Studies [IVUS]). J Am Coll Cardiol 2001;37:1478–1492. 14. Ellis SG, Vandormael MG, Cowley MJ, et al. Coronary morphologic and clinical determinants of procedural outcome with angioplasty for multivessel coronary dis- ease. Circulation 1990;82:1193–1202. 15. Brown BG, Bolson EL, Dodge HT. Quantitative computer techniques for analyzing coronary arteriograms. Prog Cardiovasc Dis 1986;28:403–418. 16. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1986;327:307–310. 17. Mudra H, di Mario C, de Jaegere P, et al. Randomized comparison of coronary stent implantation under ultrasound or angiographic guidance to reduce stent restenosis (OPTICUS Study). Circulation 2001;104:1343–1349. 18. Leber AW, Knez A, von Ziegler F, et al. Quantification of obstructive and nonob- structive coronary lesions by 64-slice computed tomography: A comparative study with quantitative coronary angiography and intravascular ultrasound. J Am Coll Cardiol 2005;46:147–154. 19. Caussin C, Larchez C, Ghostine S, et al. Comparison of coronary minimal lumen area quantification by sixty-four-slice computed tomography versus intravascular ultrasound for intermediate stenosis. Am J Cardiol 2006;98:871–876. 20. Hausleiter J, Meyer T, Hermann F, et al. Estimated radiation dose associated with cardiac CT angiography. JAMA 2009;301:500–507. 21. Maruyama T, Takada M, Hasuike Y, et al. Radiation dose reduction and coronary assessability of prospective electrocardiogram-gated computed tomography coronary angiography: Comparison with retrospective electrocardiogram-gated helical scan. J Am Coll Cardiol 2008;52:1450–1455. 22. Bischoff B, Hein F, Meyer T, et al. Impact of a reduced tube voltage on CT angiog- raphy and radiation dose: Results of the PROTECTION I study. JACC Cardiovasc Imaging 2009;2:940–946. 23. Curry SH, Lopez LM, Lambert CR, et al. Plasma concentrations and hemodynamic effects of intravenous, sublingual, and aerosolized nitroglycerin in patients undergoing cardiac catheterization. Biopharmaceutics & Drug Disposition 1993;14:107–118. 24. Pfister M, Seiler C, Fleisch M, et al. Nitrate induced coronary vasodilatation: Differ- ential effects of sublingual application by capsule or spray. Heart 1998;80:365–369.