Comparison of definitions of coronary artery reference sizes and effects on stent selection and evaluation of stent expansion

A main difference in reference methods was seen when 31 lumen-based estimations were compared to media-layer derived estimations ranging from mean 2.91 mm to mean 4.04 mm, respectively (Table 3).

The majority of identified studies recommended using a distal reference estimation for stent sizing. We found that media-based methods would result in selection of stents median 0.5 mm larger than lumen-based reference methods. Choosing the mean of the proximal and distal reference the stent selection was on average 0.75 mm larger based on media compared to lumen-based methods. Lumen size estimations neglect the regular age-related changes in vessel morphology [28]. Such frequent changes in the reference segments includes intimal thickening, positive vessel remodeling and formation of calcific, and lipid rich plaques [5]. Using angiographic visual estimation provides limited information about potential plaque burden in the reference segment or vessel wall remodeling [5]. A study by Mintz et al. showed that only 60% of 884 angiographically normal reference segments were normal when evaluated by IVUS [5]. In this regard, both angiographic visual estimation and imaging guided lumen based reference methods may frequently lead to undersized stent selection and too small expansion goals. Some clinical studies mandated to select stent size based only on the largest lumen, without evaluation of the vessel morphology [18]. A large lumen by coronary angiography could still have plaque or be a post-stenotic lumen enlargement with positive vessel remodeling [29]. Incorrect reference size estimation of more than 0.5 mm could lead to suboptimal implantation result with potential clinical impact. A particular issue is the selection of a stent platform where the maximal achievable stent dimensions are not sufficient for optimal stent expansion.

Recent imaging-guided studies aimed to overcome the limitations with lumen-based reference sizing by measurements defined by the external elastic lamina (EEL). Defining the media layer and not specifically the media-adventitia transition (EEL) was selected in this study due to a reasonable higher in-procedure feasibility and a limited error between the two definitions estimated to 0.10-0.15 mm. IVUS and OCT are in many ways similar in their capacity to define the different vessel layers, plaque types and stent dimensions. However, several key differences do exist [30]; with an image resolution of 10 μm, OCT has 10 times greater resolution compared with IVUS (20-40 μm), while the tissue penetration by OCT is limited (OCT: 1-2 mm, IVUS: 3-8 mm), in particular in presence of lipid plaques [31]. Due to the high attenuation of lipid limiting the media detection behind lipid plaque it was proposed that EEL/media estimations are only feasible using IVUS compared with OCT [10]. The ILUMIEN III study found that physicians were able to identify EEL > 180 degrees in-procedure using OCT in 84% of cases. The EEL was detected in 95% of OCT cases during subsequent corelab analysis. The higher spatial resolution of OCT in combination with required blood clearance offers a clear interface between healthy-looking segments and non-healthy looking segments [32]. The in-procedure feasibility of the individual reference methods may vary according to presence of thrombus or the level and composition of plaque in the reference segment. Such mechanisms should be clarified in future clinical evaluation of in-procedure reference size estimation.

Multiple definitions of “optimal stent expansion” have been proposed (Table 2). Stent underexpansion is an important known risk factor for stent failure. Hence, optimal and feasible evaluation of stent expansion is therefore of particular importance [6]. Expansion is assessed using either an absolute in-stent minimal stent area or diameter, or as a relative value expressed as minimal stent area/diameter to a reference size estimate.

Although expansion criteria by absolute values have been applied in several studies [15, 26, 33, 34], the strategy is infeasible when treating small vessels (< 2.5 mm) and may allow for accepting stents with expansion substantially below the reference size in large vessels. The NOBLE IVUS substudy indicated that relative rather than absolute expansion values predict prognosis in large vessels including the left main coronary artery [35]. The present study showed that evaluation of expansion as a relative value depends on the selected reference size estimation method. This includes which reference/references to use for calculation: (1) the distal or the mean of the distal and the proximal references, (2) lumen or media/EEL reference estimation or (3) area or diameter reference estimations, but also if one segment or divided segments should be used. Even with an optimal reference size, the threshold for optimal expansion is not firmly established with 80% and 90% expansion limits applied in investigated studies.

Two ongoing large-scale randomized control trials (ILUMIEN IV Trial [27] and the OCTOBER Trial [9]) are both evaluating routine OCT guidance for stenting complex coronary lesions. The studies apply different approaches for evaluation of stent expansion. Stent expansion in the ILUMIEN IV trial is defined by MSA to the post PCI lumen area. In the OCTOBER trial expansion is evaluated by MSA to the pre-PCI reference estimation. This study showed that such very important difference in definitions can result in difference in post-PCI evaluation of expansion ranging from 100 to 64% in the same case. Selecting a reference vessel segment is not trivial. The vessel lumen immediate next to the post-PCI stent could be influenced by stent overexpansion, geographical miss, plaque shift or coronary artery dissection [36, 37]. In this study ≈ 40% of cases showed the largest post-PCI OCT lumen just at the stent boarder, indicating that the edge segment closest to the stent was influenced by the stent-implantation. Using a reference measured as post-PCI stent edge lumen for evaluation of stent expansion therefore may result in a larger reference-value and a smaller expansion in comparison to the corresponding pre-PCI lumen. Numerous trials reviewed in this study did not state whether pre- or post-reference estimations were used for expansion evaluation [17, 19, 23, 24, 36, 38].

Most studies evaluate the expansion in relation to area-estimations. However, the clinical feasibility in calculating % of an area seems offhand limited and inconvenient when compared to % of diameter. Since percentages of an area cannot be transferred directly to percentages of diameter (e.g. \({0.90 x\left(\frac{d}{2}\right)}^{2}x \pi \ne\)\({\left(\frac{0.9 x d}{2}\right)}^{2}x \pi\)), physicians must be careful not to transfer cut-off values based on an area-protocol to a diameter-protocol.

Our findings indicate that reference methods (1) should be vessel and not lumen-based, (2) should be applicable to long lesions with major diameter shifts, (3) should identify reference segments more than 2 mm from the stent edge, and (4) potentially be based on diameters to provide more actionable measurements for selecting balloons sizes. The post hoc nature and limited number of cases in this study do not allow for firm conclusions. It is important to establish a future consensus based on clinical outcome data expected from ongoing randomized trials.

Retrospective analyses are inherently limited by the fact that included data was collected for a different purpose. Measurements were performed in a core laboratory setting and not during angiographic procedures. Thus, it is unknown if the applied corelab analysis resemble measurements performed quickly during a procedure. The theoretic choice of stent size was performed according to established criteria but could differ from decisions made by treating physicians. Matching of pre - and post -PCI OCT-scans is limited by cardiac motion artifacts known to occur in 20–30% of cases [39]. To reduce this inherent limitation, visual confirmation of matched cross sections was performed by an experienced OCT observer.