Validation of 4 French Catheters for Quantitative Coronary Analysis: In Vivo Variability Assessment Using 6 French Guiding Cathe

ABSTRACT: Although numerous studies have established the utility of 4 French (Fr) catheters for routine coronary angiography, its adequacy for automatic quantitative coronary analysis has not been previously assessed. Methods. In 32 consecutive patients, coronary angiography was performed sequentially with 4 Fr diagnostic catheters and 6 Fr guiding catheters after intracoronary nitroglycerin. A total of 43 lesions were evaluated for quantitative analysis using both types of catheter as scaling devices. Possible differences in the reference diameter, minimal luminal diameter and percent diameter stenosis were evaluated. All measurements were performed offline by the same operator and intraobserver variability estimation was performed by repeating the evaluation in 12 lesions randomly selected after 1 month. Results. The mean reference diameter was 2.98 ± 0.48 mm, mean minimal luminal diameter was 1.00 ± 0.52 mm and percent diameter stenosis was 67.1 ± 15.3%. Accuracy (mean difference of values) was -0.009 mm for reference diameter, 0.005 mm for minimal luminal diameter and -0.25% for percent diameter stenosis. Precision (mean standard deviation of the differences) was 0.17 mm for reference diameter, 0.19 mm for minimal luminal diameter and 5.93% for percent stenosis. Linear correlation for these three variables was 0.94, 0.93 and 0.93, respectively. Intraobserver variability analysis showed similar values for accuracy, precision and linear correlation. Conclusions. Angiography with 4 Fr catheters allows adequate quantification of luminal diameters as compared to most accepted clinical standards. These results may have implications for the selection of diagnostic catheters for routine follow-up assessment of percutaneous coronary interventions.

Key words: catheterization, coronary artery disease, quantitative coronary angiography

Quantitative coronary angiography (QCA) has gained widespread use in routine practice of cardiac catheterization units for both research and clinical applications. Since the introduction and development of the technique more than a decade ago,^1,2 considerable improvements have been made in x-ray image acquisition and storage and also in catheter design that have imposed changes in routine practice in many centers. Among these, the miniaturization of diagnostic and guiding catheters is becoming a major accepted means of reducing overall procedural complications, especially those related to the vascular access site.^3–7 Downsizing diagnostic and angioplasty catheters has also facilitated diffusion of transradial procedures that shorten time to deambulation and reduce bleeding complications.⁸ Typically, diagnostic catheters used during transradial procedures are 4 or 5 French (Fr) in size. However, although widely accepted for diagnostic purposes, the validity of these “ultra-small” catheters for QCA is uncertain. Since the reliability of QCA measurements depends on good-quality imaging that includes correct filling of coronary arteries and good visualization of the catheter for calibration, the use of smaller diagnostic catheters needs a critical analysis of suitability and variability that has not yet been performed. The present study was designed to evaluate 4 Fr diagnostic catheters for QCA measurements in vivo, using 6 Fr conventional guiding catheters as reference scaling devices. Methods Patients and study protocol. Thirty-two consecutive patients constituted the study group. Exclusion criteria were history of renal failure, moderate to severe left ventricular dysfunction, acute ST elevation myocardial infarction or ongoing chest pain, total coronary occlusions and suboptimal visualization of coronary arteries due to obesity or incomplete coronary filling. The study was approved by the local Institutional Review Board and all patients gave written informed consent before inclusion. A total of 43 lesions were selected for QCA analysis and were included in the variability assessment. Patients were selected after routine diagnostic procedures performed with commercially available 4 Fr catheters. After the decision for coronary intervention was made, either if performed ad hoc or as a scheduled procedure, a 4 Fr diagnostic catheter was advanced for target coronary vessel selective cannulation. Approximately 60 seconds after administering 200 µg of intracoronary nitroglycerin, a minimum of two orthogonal views were recorded. Immediately after 4 Fr imaging, the same projections were filmed with the 6 Fr guiding catheter before advancement of the guidewire. After baseline imaging, coronary angioplasty and post-procedural management were performed as usual. The procedures were performed by four different operators. QCA analysis was performed offline by a different investigator (RC). One month after the initial QCA measurements, the same investigator performed a second analysis of the angiographic recordings in 12 cases randomly selected for intraobserver variability evaluation. This second analysis was performed without knowledge of previous results or of the frames selected in the first evaluation. Coronary angiography. The angiographic equipment used was a General Electric Advantx (GE Medical Systems, Paris, France). The system allows direct digital acquisition and storage on DICOM format. Images for QCA analysis were recorded on a 16 cm intensifier at 25 frames/second. All study cases were recorded on CD-ROM for offline review and quantitative measurements. Coronary angiography was performed with 4 Fr diagnostic catheters (Infinity™, Cordis Corporation, Miami, Florida or Outlook™, Terumo, Leuven, Belgium) and with 6 Fr guiding catheters (either Vistabrite tip, Cordis Corporation, Miami, Florida or Z2, Medtronic Inc., Minneapolis, Minnesota), as specified per protocol. Low-osmolality, nonionic contrast agent (iohexol; Omnitrast 300® Schering AG, Germany) was used in all cases. The empty catheter (not filled with contrast agent) was recorded in an approximately centered position and used for calibration. Stenosis quantification was performed in diastolic frames with completely opacified segments that showed the least foreshortening and no sidebranch overlap. Two views were taken for each coronary artery, with at least 60° different angulation, and the best image was selected. QCA measurements. QCA measurements were performed with a validated software package available by General Electric GEMnet equipment, and described in detail elsewhere.^9,10 Briefly, the operator defines the region of interest by marking an initial and a final point within the catheter or vessel contour. After automatic centerline delineation, the edges are defined using oriented monodimensional first derivative. Subsequently, the centerline is redefined and vessel diameters are computed perpendicularly. Calibration for this study was obtained using the external diameter of the catheter image as the scaling device. In order to enhance clinical applicability of the present results, no attempt of scaling catheter actual dimensions with precision micrometers was made. Rather, we preferred conventional software estimations of the catheters, determined by labeled external size in French units. QCA variables used in this study were reference vessel diameter (RD); minimal luminal diameter (MLD) and percent diameter stenosis (%DS). As explained, catheter and vessel contours were automatically determined by the software and no manual correction was attempted. Statistics. Continuous variables are represented as means and standard deviations. Variability between anatomic measurements extracted from QCA analysis with 4 and 6 Fr angiography was estimated by calculating accuracy (the signed mean difference between the two measurements), precision (standard deviation of differences), coefficient of variation (standard deviation of differences divided by the mean obtained with the 6 Fr catheter) and correlation coefficient. Intraobserver variability was measured by calculating the same parameters (accuracy, precision, coefficient of variation and correlation coefficient) for 4 Fr QCA and 6 Fr QCA obtained by the same investigator after a 1-month interval. Results A total of 43 lesions were analyzed. Eighteen lesions were located in the left anterior descending artery, nine in the circumflex artery and 16 in the right coronary artery. Comparisons between 4 and 6 Fr determined QCA data are presented in Table 1. As designed by the study protocol, a wide range of lesion severity was used (minimal stenosis was 15.3% and maximal stenosis was 95%); however, the vast majority were angiographically severe (mean %DS, 67.1 ± 15.3%). Also, a wide range of vessel dimensions was included, with maximal RD of 4.3 mm and minimal RD of 2.0 mm. Values of accuracy for the 3 QCA variables were close to zero (-0.009 mm, 0.005 mm and -0.25% for RD, MLD and %DS, respectively). Calculation of MLD implied a higher coefficient of variation (19.2%). An excellent correlation was found between values determined with the 2 sizes of catheter (Table 1 and Figure 1). Variability of measurements was well distributed among all ranges of diameter and lesion severity, without significant prevalence of positive or negative values (Figure 2). Intraobserver variability. Measurements in 12 randomly selected lesions were repeated by the same investigator without recording the baseline frame selected for estimation of reproducibility in QCA analysis with both sizes of catheter. Variability measurements are shown in Table 2. Values of precision and coefficient of variation were quite similar to those obtained for comparisons of 4 and 6 Fr angiography. Discussion The main finding of the present study is that QCA measurements with 4 Fr diagnostic catheters are similar to those obtained with 6 Fr guiding catheters, especially for determination of reference diameters, with no significant bias related to lumen size, and have good reproducibility. The small degree of variation found is expected for most QCA software packages currently available and is acceptable for clinical application and probably also for investigational purposes. Potential bias of QCA analysis. The analysis of coronary diameters by QCA is essential both for clinical practice and research. QCA allows reduction of inter- and intraobserver variability when compared to conventional visual estimation.^1,2 However, it is important to understand that quantitative angiography measurements in commercially available software packages are subject to some degree of variability that depends on several factors.¹¹ For in vivo assessment, if changes in vasomotor tone are diminished by systematic administration of nitroglycerin, the most important sources of inaccuracy are operator-dependent variables. These include comparison of different angiographic projections, poor contrast filling and selection of different frames or different coronary reference segments. Calibration is also an important source of error. The catheter’s nontapered tip is used as the scaling device in standard practice. Since correct calibration is a critical step in calculating coronary dimensions, estimation of variability among currently available catheter sizes is important for understanding and interpretation of results of longitudinal studies in interventional cardiology. Currently, the majority of interventional cardiology units use 6 Fr catheters for both diagnostic and angioplasty procedures and this is the common recommended size for follow-up angiographic studies in clinical research. However, smaller catheters such as 5 and 4 Fr are gaining popularity. Critics of this “mini-invasive” approach for coronary angiography refer to poor contrast filling, streaming effects and suboptimal visualization of the catheter as important limitations of 4 Fr angiography. However, several studies, including one multicenter randomized trial comparing 4 and 6 Fr manual injection, suggest similar quality with a trend to less contrast agent requirement with the smaller catheters.^5–7 Although suitability of 4 Fr catheters for routine coronary angiography is becoming clear, to our knowledge this is the first prospective study that has validated 4 Fr angiography for QCA applications as well. From a hypothetical standpoint, the use of smaller scaling devices implies an intrinsic inaccuracy due to the greater pixel/catheter ratio. Consequently, minimal errors in determination of the scaling factor are magnified when estimating vessel diameters. The impact of this potential source of error is addressed in the present study. Since we have used the same equipment and QCA software, and tried to minimize possible differences in angiographic settings by sequential selective coronary injections and analysis of projections with exactly the same angulation, the small differences between measurements found in this study should be interpreted as a consequence of variation in the calibration process or intrinsic variability of the software. Variability between 4 and 6 Fr QCA. Originally, most studies based on QCA observations were performed with 7 and 8 Fr angiography. Curiously, although 6 Fr catheters are now the most widely accepted for diagnostic and interventional purposes, QCA validation of 6 Fr derived data is less extensive. Legrand et al.¹² performed a retrospective comparison of QCA data obtained with 6 Fr diagnostic and 8 Fr guiding catheters. They found that accuracy and precision were -0.036 mm and 0.22 mm for reference diameter and 0.027 mm and 0.19 mm for minimal diameter, respectively. Interobserver and intraobserver reproducibility were greater for 8 Fr angiography, especially for the reference diameter. The authors concluded that 6 Fr angiography is less reliable than 8 Fr angiography and is probably not useful for serial studies looking for progression/regression of atherosclerosis and proper estimation of luminal loss following angioplasty. Despite this, most QCA core laboratories now accept 6 Fr angiography for baseline and follow-up evaluation of investigational devices. A similar degree of variability was shown in the present study for 4 and 6 Fr angiography, although intraobserver differences are smaller than those reported by Legrand et al. Methodological issues could explain the better performance of 4 and 6 Fr catheters in our study, since due to the retrospective nature of the study of Legrand et al., evaluation of different angiographic angulations was common and systematic intracoronary nitroglycerin was not administered. Besides the comparison with previous data obtained with 6 and 8 Fr devices, it is also important to consider the reproducibility that may be expected from commercially available QCA systems. Experimental in vivo validation of the Cardiovascular Angiography Analysis System (CAAS II) revealed an accuracy of -0.01 mm and precision of 0.18 mm for estimation of obstruction diameter.¹³ Similarly, Reiber et al.¹⁴ found a variability of 0.18 mm for minimal diameter and 5.28% for obstruction stenosis with the Phillips DCI-SX system. Interobserver reproducibility study of coronary estimations with the QCA system used in the present work (General Electric QA package) rendered a precision of 0.19 mm for the reference diameter.¹⁵ Thus, currently worldwide-accepted QCA systems present a similar degree of variability compared to the present work with 4 Fr angiography. Importantly, observed differences in coronary measurements obtained here with 4 Fr diagnostic catheters by the same user (intraobserver variability) are very similar to those found with 6 Fr guiding catheter angiography, meaning that the luminal diameters measured with 4 Fr are highly reproducible. Study limitations. The present study represents an analysis of suitability of 4 Fr catheters by comparison with clinically widely accepted 6 Fr guiding catheter angiography. Accordingly, we have not performed an in vitro assessment of the adequacy of 4 Fr angiography for determination of actual intraluminal diameters. In addition, we have no data regarding interobserver reproducibility. However, the most common source of interobserver variability is selection of different coronary segments, and not the calibration process. In this regard, the analysis of interobserver variability would most likely not change the present conclusions. Conclusions. QCA estimation of coronary dimensions and severity of coronary stenosis using 4 Fr diagnostic catheters as scaling devices results in luminal diameters similar to the more widely accepted 6 Fr guiding catheter angiography. The reported values of accuracy and precision, as well as intraobserver reproducibility, are equivalent to the accepted standards of currently used QCA software packages. Thus, provided that good-quality images can be obtained, 4 Fr catheters are probably appropriate when quantitative angiography is a necessary part of the diagnostic study.