Abstract
Purpose
Frequency-domain optical coherence tomography (FD-OCT) is a high-resolution imaging tool (~ 10–15 μm), which enables near-histological in-vivo images of the coronary vessel wall. The use of the technique is increasing, both for research- and clinical purposes. This study sought to investigate the intra- and interobserver reliability, as well as the intra-catheter reproducibility of quantitative FD-OCT-assessment of morphometric stent parameters and qualitative FD-OCT-evaluation of strut coverage in 10 randomly selected 6-month follow-up Nobori® biolimus-eluting stents (N-BESs).
Methods
Ten N-BESs (213 cross sectional areas (CSAs) and 1897 struts) imaged with OCT 6 months post-implantation were randomly selected and analyzed by 2 experienced analysts, and the same 10 N-BESs were analyzed by one of the analysts 3 months later. Further, 2 consecutive pullbacks randomly performed in another 10 N-BESs (219 CSAs and 1860 struts) were independently assessed by one of the analysts.
Results
The intraobserver variability with regard to relative difference of mean luminal area and mean stent area at the CSA-level was very low: 0.1% ± 1.4% and 0.5% ± 3.2%. Interobserver variability also proved to be low: − 2.1% ± 3.3% and 2.1% ± 4.6%, and moreover, very restricted intra-catheter variation was observed: 0.02% ± 6.8% and − 0.18% ± 5.2%. The intraobserver-, interobserver- and intra-catheter reliability for the qualitative evaluation of strut coverage was found to be: kappa (κ) = 0.91 (95% confidence interval (CI): 0.88–0.93, p < 0.01), κ = 0.88 (95% CI: 0.85–0.91, p < 0.01), and κ = 0.73 (95% CI: 0.68–0.78, p < 0.01), respectively.
Conclusions
FD-OCT is a reproducible and reliable imaging tool for quantitative evaluation of stented coronary segments, and for qualitative assessment of strut coverage.
Highlights
- •
Frequency-domain optical coherence tomography (FD-OCT) is increasingly adopted in the catherization laboratories.
- •
This manuscript adresses the intraobserver, the interobserver, and the intra-catheter reproducibility of FD-OCT.
- •
Morphometric parameters within stented coronary segments can be reproduced with high accuracy at different time points, between different analysts, and between different pullbacks.
- •
Qualitative assessment of strut coverage by FD-OCT has excellent intra- and interobserver and good intra-catheter agreement.
1
Introduction
Frequency-domain optical coherence tomography (FD-OCT) is a high-resolution intravascular imaging modality that produces near-histological in-vivo images of the coronary vessel wall with very fast data acquisition . This tomographic technique has revolutionized the field of intracoronary imaging, and the interest in using the technology is increasing, both for clinical purposes in guiding and evaluating stent implantations, and as a research tool . The validity of the technique and acquired results is highly dependent on its reproducibility.
In previous studies , intraobserver-, interobserver- and intra-catheter reproducibility for morphometric stent parameters and quantitative measurements of neointimal hyperplasia (NIH) thickness have been found to be reliable. However, studies exclusively devoted to investigate the reproducibility of FD-OCT-assessed strut coverage at different time points, between different analysts , and between different pullbacks are very limited.
Deficient strut coverage has been documented as an important morphometric predisposing factor for later thrombotic events following drug-eluting stent (DES) implantation in previous ex- and in-vivo studies . OCT offers a distinctive value to determine the presence of strut coverage , explaining why OCT-assessed strut level-analysis represents a surrogate marker of DES safety in many recent and ongoing clinical trials. Currently, there is broad agreement on considering OCT-detected strut coverage as a categorical variable (covered or uncovered), however a complete standardized approach is lacking .
OCT is progressively adopted in the catheterization laboratories and involved in the clinical decision-making, and several OCT-driven clinical trials using strut coverage as a surrogate endpoint are yet to come, justifying further reproducibility studies regarding the technology.
The aim of this study was therefore to investigate the validity of the FD-OCT imaging technique by evaluating: (1) the intraobserver, (2) the interobserver, and (3) the intra-catheter reliability for quantitative assessment of stented coronary segments, and qualitative assessment of strut coverage.
2
Methods
2.1
Study population
Eighty-five previously Nobori® biolimus-eluting stented coronary segments (from 85 asymptomatic patients undergoing 6-month repeat OCT) were available for analysis. Ten of these 6-month follow-up pullbacks (213 cross sectional areas (CSAs) and 1897 struts) were randomly selected by a computerized randomization system.
Further, in another 10 coronary segments (219 CSAs and 1860 struts), two pullbacks had been randomly performed at the 6-month follow-up.
The institutional review board of our institution approved the study, and written informed consent was obtained from all patients.
2.2
FD-OCT imaging system
OCT imaging of the stented segments was performed with a frequency-domain OCT system (C7-XR™ OCT imaging or Ilumien™ PCI Optimization OCT system; LightLab Imaging, Inc., St. Jude Medical, St. Paul, MN, USA). The imaging was performed following administration of 200 μg intracoronary nitroglycerin and a 5000 IU unfractionated heparin dose. A Z-offset calibration was performed manually prior to introduction of the imaging catheter into the target vessel. Stent edges were used as landmarks for the region of interest. A 2.7 Fr C7 Dragonfly™ Imaging Catheter (LightLab Imaging, Inc., St. Jude Medical, St. Paul, MN, USA), flushed with 20 ml undiluted contrast was used. Contrast medium was flushed manually and continuously (rate of 4–5 ml/s for 3–4 s) through the guiding catheter during image acquisition. Motorized pullback OCT imaging was performed at a pullback rate of 20 mm/s throughout the stent, and the maximal pullback length allowed by the system was 54 mm.
In 10 randomly chosen patients, an additional image acquisition was performed under equivalent conditions, using the same imaging catheter. The double pullbacks were performed in fast succession (with pullback 2 acquired approximately 30 s after pullback 1), after re-engagement of the imaging catheter.
All images were digitally stored in the FD-OCT system console and on DVD for later off-line analysis.
2.3
FD-OCT analysis
Quantitative OCT analysis was performed using the LightLab OCT proprietary software (Offline Review Workstation, LightLab Imaging, Inc., St. Jude Medical, St. Paul, MN, USA). Optimal calibration was secured prior to initiation of the analysis, and if necessary, Z-offset was adjusted.
Lesions were analyzed at the cross sectional level with an interval of 1 mm (every 5 frames). The target lesion was defined as the stented segment between the first and last frame where struts were visible.
Observer and intra-catheter agreements for the following morphometric stent parameters were evaluated: lumen and stent areas were traced in all CSAs. Lumen contours were obtained automatically by the software system, and additional manual corrections were performed when required. The minimal luminal area (MLA) was defined as the smallest luminal CSA in the measured luminal CSAs within the stented segment. The stent CSA comprised the area bounded by the stent border. Stent contours were traced using a multiple point detection function, where support points were applied in the middle of the endoluminal border of each stent strut, and a semi-automated contour was then utilized linking the points. The minimal stent area (MSA) comprised the smallest stent CSA in the measured stent CSAs within the stented segment. Mean lumen- and stent CSAs were calculated for each stented segment by adding the traced CSAs and dividing by the total number of CSAs analyzed.
Strut coverage was assessed using the following healing subtypes: 1) definitely uncovered, 2) uncovered/fibrin, 3) partially uncovered, 4) covered protruding, 5) covered embedded, and 6) covered proliferative. Subtypes 1, 2 and 3 were deemed uncovered, while subtypes 4, 5 and 6 were categorized as covered. Fig. 1 illustrates the various healing subtypes. Subtypes 1, 2 and 3 were defined as having an NIH thickness of 0 μm. In subtypes 4, 5 and 6 strut coverage was measured as the distance between the endoluminal side of the strut from the midpoint of its long axis and the intersection of the lumen contour with a straight line between the endoluminal side of the strut and the gravitational center of the vessel . There were no zoom-setting restrictions in the three subanalyses, and both analysts A and B evaluated individual/single stent struts at 0.1 mm.
2.4
Assessment of intraobserver, interobserver and intra-catheter reproducibility
To assess the intraobserver reliability, 10 pullbacks (213 CSAs and 1897 struts) were analyzed by one experienced OCT-analyst (analyst A) at time 0 (defined as time point A1), and re-analyzed 3 months later (defined as time point A2).
To evaluate interobserver reliability, 2 experienced OCT-analysts (analyst A and analyst B) analyzed the same 10 pullbacks (213 CSAs and 1897 struts) independently of each other.
Analyst A also analyzed another 10 double pullbacks (pullback 1 and pullback 2), comprising 219 CSAs and 1860 struts, independently of each other. The various variabilies are illustrated in Fig. 2 .
2.5
Statistical analysis
The intra- and interobserver reliability, and intra-catheter reproducibility for selected morphometric parameters within the stented segments were tested at (1) the stent level, and at (2) the CSA level. Values are expressed as mean ± SD. Further, the absolute and the relative differences (absolute difference divided by the average) are provided. Differences in luminal and stent areas at the CSA level are also expressed in scatter plots and Bland–Altman plots . The limits of agreement were calculated as mean difference ± 1.96 SD.
Kappa statistics (κ) was performed to determine consistency between different time points (analyst A (time point A1 versus time point A2)), different observers (analyst A versus analyst B), and different pullbacks (pullback 1 versus pullback 2) on the number of struts evaluated as covered.
A two-sided p-value of < 0.05 is considered statistically significant.
Statistical analysis was performed using SPSS version 22.0 (SPSS, Chicago, IL, USA).
2
Methods
2.1
Study population
Eighty-five previously Nobori® biolimus-eluting stented coronary segments (from 85 asymptomatic patients undergoing 6-month repeat OCT) were available for analysis. Ten of these 6-month follow-up pullbacks (213 cross sectional areas (CSAs) and 1897 struts) were randomly selected by a computerized randomization system.
Further, in another 10 coronary segments (219 CSAs and 1860 struts), two pullbacks had been randomly performed at the 6-month follow-up.
The institutional review board of our institution approved the study, and written informed consent was obtained from all patients.
2.2
FD-OCT imaging system
OCT imaging of the stented segments was performed with a frequency-domain OCT system (C7-XR™ OCT imaging or Ilumien™ PCI Optimization OCT system; LightLab Imaging, Inc., St. Jude Medical, St. Paul, MN, USA). The imaging was performed following administration of 200 μg intracoronary nitroglycerin and a 5000 IU unfractionated heparin dose. A Z-offset calibration was performed manually prior to introduction of the imaging catheter into the target vessel. Stent edges were used as landmarks for the region of interest. A 2.7 Fr C7 Dragonfly™ Imaging Catheter (LightLab Imaging, Inc., St. Jude Medical, St. Paul, MN, USA), flushed with 20 ml undiluted contrast was used. Contrast medium was flushed manually and continuously (rate of 4–5 ml/s for 3–4 s) through the guiding catheter during image acquisition. Motorized pullback OCT imaging was performed at a pullback rate of 20 mm/s throughout the stent, and the maximal pullback length allowed by the system was 54 mm.
In 10 randomly chosen patients, an additional image acquisition was performed under equivalent conditions, using the same imaging catheter. The double pullbacks were performed in fast succession (with pullback 2 acquired approximately 30 s after pullback 1), after re-engagement of the imaging catheter.
All images were digitally stored in the FD-OCT system console and on DVD for later off-line analysis.
2.3
FD-OCT analysis
Quantitative OCT analysis was performed using the LightLab OCT proprietary software (Offline Review Workstation, LightLab Imaging, Inc., St. Jude Medical, St. Paul, MN, USA). Optimal calibration was secured prior to initiation of the analysis, and if necessary, Z-offset was adjusted.
Lesions were analyzed at the cross sectional level with an interval of 1 mm (every 5 frames). The target lesion was defined as the stented segment between the first and last frame where struts were visible.
Observer and intra-catheter agreements for the following morphometric stent parameters were evaluated: lumen and stent areas were traced in all CSAs. Lumen contours were obtained automatically by the software system, and additional manual corrections were performed when required. The minimal luminal area (MLA) was defined as the smallest luminal CSA in the measured luminal CSAs within the stented segment. The stent CSA comprised the area bounded by the stent border. Stent contours were traced using a multiple point detection function, where support points were applied in the middle of the endoluminal border of each stent strut, and a semi-automated contour was then utilized linking the points. The minimal stent area (MSA) comprised the smallest stent CSA in the measured stent CSAs within the stented segment. Mean lumen- and stent CSAs were calculated for each stented segment by adding the traced CSAs and dividing by the total number of CSAs analyzed.
Strut coverage was assessed using the following healing subtypes: 1) definitely uncovered, 2) uncovered/fibrin, 3) partially uncovered, 4) covered protruding, 5) covered embedded, and 6) covered proliferative. Subtypes 1, 2 and 3 were deemed uncovered, while subtypes 4, 5 and 6 were categorized as covered. Fig. 1 illustrates the various healing subtypes. Subtypes 1, 2 and 3 were defined as having an NIH thickness of 0 μm. In subtypes 4, 5 and 6 strut coverage was measured as the distance between the endoluminal side of the strut from the midpoint of its long axis and the intersection of the lumen contour with a straight line between the endoluminal side of the strut and the gravitational center of the vessel . There were no zoom-setting restrictions in the three subanalyses, and both analysts A and B evaluated individual/single stent struts at 0.1 mm.
2.4
Assessment of intraobserver, interobserver and intra-catheter reproducibility
To assess the intraobserver reliability, 10 pullbacks (213 CSAs and 1897 struts) were analyzed by one experienced OCT-analyst (analyst A) at time 0 (defined as time point A1), and re-analyzed 3 months later (defined as time point A2).
To evaluate interobserver reliability, 2 experienced OCT-analysts (analyst A and analyst B) analyzed the same 10 pullbacks (213 CSAs and 1897 struts) independently of each other.
Analyst A also analyzed another 10 double pullbacks (pullback 1 and pullback 2), comprising 219 CSAs and 1860 struts, independently of each other. The various variabilies are illustrated in Fig. 2 .
