Feasibility of Frequency-Domain Optical Coherence Tomography for Coronary Stent Imaging: A Matter of Definitions

To the Editor:

With great interest, we read the article “Feasibility and safety of the second generation, frequency domain optical coherence tomography (FD-OCT): a multicenter study” by Yoon et al.¹ In fact, the authors are to be congratulated for providing this timely report, based on FD-OCT: the state-of-the-art technology of intravascular coronary imaging. Their landmark conclusion was that “the new second-generation FD-OCT system provides fast and reliable resolution images of the coronary artery. The pullback can be safely performed over long segments of the artery without serious adverse events.”

Despite being clearly safe in terms of the peri- and immediate postprocedural complications, I am quite concerned about the feasibility of the technique as described in the report. First, the primary endpoint to demonstrate diagnostic effectiveness was the achievement of a clear-image length (CIL) of more than 24 mm. The assignment of the 24 mm cutoff point was based on a meta-analysis of intravascular ultrasonography substudies from the TAXUS IV, V, and VI trials demonstrating that 90% of stents have a length less than 24 mm.² This meta-analysis necessarily included patients from the TAXUS IV trial with a mean stent length of 18.5 ± 5.0 mm. Moreover, CIL was measured as the cumulative sum of sporadically distributed clear-image frames, which were not essentially consecutive, nor continuous over one compact segment. Hence, achievement of 24 mm CIL may not necessarily correspond to the index stent region.

Additionally, considering that the pullback scanned, in essence, 50 mm of the vessel length, achievement of 24 mm CIL (less than 50% of the scanned length) in 94% of the cases would not be sufficient to decide about feasibility. When exploring the  effectiveness of FD-OCT for imaging intracoronary stents, a primary endpoint focused on the index stent would be adopted, rather than one concerning an absolute summation of clear-image “millimeters” within the scanned region. Second, a second attempt was needed in 5 patients (10.6%) due to suboptimal image quality of the first attempt. This would eventually translate into a first-attempt feasibility of 83.4%.

Out of these second attempts in 5 patients, clear stent length was not measurable in 1, although full-length CIL was obtained in all 5 patients during the second attempt, further raising concern about the dissociation between the primary endpoint and the stent to be evaluated. Third, when assessing the strut malapposition, no correction was made for half the blooming of the strut artifact, which makes it very liable to overestimate the malapposition distance. No wonder then that 6.8% of the struts were assigned as malapposed. Finally, the 50-patient cohort of the study was collected from 3 centers over a period of 8 months. This would give an enrollment rate of 2 patients per month in each center, which makes enrollment quite vulnerable to selection bias. Furthermore, the long list of exclusion criteria, including both patient and lesion characteristics, makes the cohort far from reflecting real-life clinical practice, a prime goal for a feasibility study.

References

1. Yoon JH, Di Vito L, Moses JW, et al. Feasibility and safety of the second-generation, frequency domain optical coherence tomography (FD-OCT): a multicenter study. J Invasive Cardiol. 2012;24(5):206-209.
2. Escolar E, Mintz GS, Popma J, et al. Meta-analysis of angiographic versus intravascular ultrasound parameters of drug-eluting stent efficacy (from TAXUS IV, V, and VI). Am J Cardiol. 2007;100(4):621-626. 

Sincerely,

Wail Nammas, MD 

Faculty of Medicine, Ain Shams University 
Cardiology Department, Abbassia Cairo, Egypt
Email: wnammas@hotmail.com

In Reply:

Dear Dr Nammas,

We also recognize a single criterion of merely achieving the 24 mm CIL might not be sufficient to satisfy the feasibility. However, frame drop-out in the middle of pullback was extremely rare. Also, since there are 5 frames in 1 mm, skipping a couple of frames rarely affect stent evaluation in the clinical setting. In our study, most blood interference occurred at the end of pullback. In addition, in the result section of the article, we have further clarified that the procedure success rate of CIL >24 mm was 94.0% in our study by providing the actual median value of CIL of 43.2 mm (83.9% of the total length of image of 51.5 mm).

The cumulative summation of clear-image frames (CIF) does not assume the frames to be consecutive without interruption. We agree that theoretically the achievement of 24 mm CIL might not correspond to the index stent region of interest out of 50 mm pullback length. However, as the actual median length of CIL (43.2 mm) has reflected, most of the index stent areas within the pullback image were visualized as clear image. Like any other study attempting to use a new device, there was a learning curve. Once the operator becomes familiar with the new catheter system, especially regarding the guiding catheter position, the need for a second pullback rapidly drops and a 20 mm stent could be visualized in the vast majority of cases. In the discussion section, we have shown that the average actual stent length in our samples was 16.7 ± 3.7 mm, and the average stented area with clear image in our first pullback was 15.9 ± 4.4 mm, showing 95.4% of stent catch rate.

In assessing strut malapposition, the image analysis was done by an independent core lab. However, two different methods are currently used, as reviewed by Mehanna et al.¹ The first method is to measure the distance from the center of the blooming to the vessel wall. Another method consists of starting from the inner surface of the blooming to the vessel wall, and correcting for half of the thickness of the blooming (usually 18 µm). The first method was adopted for this analysis. In addition, the percentage of malapposed strut obtained in our study was close to the percentage reported in previous works considering the variability associated with stent type (3.55 ± 5.16% for paclitaxel-eluting stent,² 11.6 ± 6.6% for sirolimus-eluting stent vs 3.7 ± 5.2% for paclitaxel-eluting stent.³

This was the first feasibility and safety study. However, the exclusion criteria used in our study have been commonly used in the subsequent OCT studies.^4,5 The use of lesion-specific exclusion criteria of FD-OCT was due to its well-known technical limitation in assessing large vessel or ostial lesion, and to avoid any bias linked to stent sizes.

References

1. Mehanna EA, Attizzani GF, Kyono H, et al. Assessment of coronary stent by optical coherence tomography, methodology and definitions. Int J Cardiovasc Imaging. 2011 Feb;27(2):259-69.
2. Guagliumi G, Bezerra HG, Sirbu V, et al. Serial assessment of coronary artery response to paclitaxel-eluting stents using optical coherence tomography. Circ Cardiovasc Interv. 2012 Feb 1;5(1):30-8.
3. Tanigawa J, Barlis P, Dimopoulos K, et al. The influence of strut thickness and cell design on immediate apposition of drug-eluting stents assessed by optical coherence tomography. Int J Cardiol. 2009 May 15;134(2):180-8.
4. Yonetsu T, Kakuta T, Lee T, et al. Impact of plaque morphology on creatine kinase-MB elevation in patients with elective stent implantation. Int J Cardiol. 2011 Jan 7;146(1):80-5.
5. Porto I, Di Vito L, Burzotta F, et al. Predictors of periprocedural (type IVa) myocardial infarction, as assessed by frequency-domain optical coherence tomography. Circ Cardiovasc Interv. 2012 Feb 1;5(1):89-96, S1-6. 

Sincerely,

Joo Heung Yoon, MD

Research Fellow, Division of Cardiology
Beth Israel Deaconess Medical Center,
Harvard Medical School
Boston, Massachusetts

Luca Di Vito, MD, PhD

Instructor, Interventional Cardiology
San Giovanni Hospital
Rome, Italy

Ik-Kyung Jang, MD, PhD

Professor of Medicine, Department of Medicine
Massachusetts General Hospital,
Harvard Medical School
Boston, Massachusetts