Bifurcation Coronary Artery Disease: Current Techniques and
Future Directions (Part 1)

Published randomized trials of drug-eluting stents (DES) have shown reduced rates of restenosis and target lesion revascularization (TLR) in noncomplex lesions compared to bare-metal stents (BMS).¹ Given this documented success of the antiproliferative therapies in relatively simple lesions, interventional cardiologists have developed techniques and devices hoping to demonstrate similarly durable results in more complex lesion subsets.^2–6 Bifurcation coronary artery disease (CAD) comprises an important subset of lesions inadequately studied and thus far unapproved for DES use. The optimal management strategy for this complex set of lesions continues to be the focus of intense interventional research and debate.^7,8

The results of recent bifurcation trials have been disappointing, irrespective of the stent platform or dedicated device used. Relative to percutaneous revascularization of non-bifurcated lesions, higher restenosis rates and unacceptable procedural complications have characterized the bifurcation studies.^2,7,9 One of the most frequently cited procedural shortcomings has been the inability to adequately cover the side branch ostium. Depicted angiographically (Figure 1) and also with intravascular ultrasound (IVUS) (Figure 2), this problem has led to an increased incidence of ostial side branch restenosis and subsequent TLR.¹⁰ In addition to gaps in stent coverage at the side branch ostium, the success of current techniques has been limited by such angiographic predictors of failure as stent distortion and redundancy of metal at the carina of the lesion (Figure 3). Both of these have been implicated in the need for TLR, and in the increased incidence of stent thrombosis.^7,11

The following is a review of those strategies currently being employed in the management of bifurcation CAD. In addition, we will review those dedicated bifurcation devices currently in developmental and testing stages. Bifurcation disease remains a tremendous challenge. As this review will show, success in this arena remains elusive, but the potential for success appears great.
Morphology of the bifurcation. Bifurcation lesion morphology, quite simply, is complex. The severity of the disease in the main branch and the side branch(es), the absolute and relative diameters of the involved vessels, the amount and distribution of calcium and fibrous tissue in the lesion, and the angle of the bifurcation are only some of the factors that have a direct bearing on the outcome of revascularization efforts.⁷
There have been multiple classifications proposed to morphologically distinguish bifurcation coronary lesions over the past several years, primarily based on the presence of disease in the main branch alone, side branch alone, or both.^12–15 Each classification scheme differs only slightly in how it describes the presence of disease in the main branch proximal and/or distal to the level of the carina, as well as disease in the side branch ostium. It has been widely held that certain lesion characteristics may predict treatment success using currently accepted techniques and DES platforms. Despite this assertion, none of these widely used classification schemes, based solely on anatomic distribution of disease, has been proven to be sufficiently predictive of procedural success. Any successful treatment strategy for bifurcation lesions must factor in a wide variety of anatomic considerations. While the distribution of disease in the main and side branch vessels is critical, so to are issues of side branch angulation, extent of lesion calcification and fibrous tissue buildup, as well as vessel diameter. If one likens the heterogeneity of bifurcation coronary lesions to that seen with fingerprints, it becomes clearer why the one-size-fits-all approach is not appropriate for bifurcation revascularization techniques or for dedicated devices; no single strategy has been shown to suffice.⁴
While certain lesion characteristics are associated with better revascularization results, the nonuniformity of bifurcationlesions has made it impossible to reliably classify lesion types in any meaningful way with regard to expected outcomes. The ACC/AHA Lesion Classification System, commonly used for describing CAD, does not apply.^16–20 Of several bifurcation classifications published in the literature, the Lefevre system¹² is widely recognized. This system focuses primarily on describing the basic anatomic patterns of plaque distribution (Figure 4). The categories described, however, have no meaningful correlation with outcomes. Of note, most disease patterns described by Lefevre are not “true bifurcation” lesions (Lefevre Type 1) at the time of diagnostic angiography, but instead may degenerate into bifurcation disease once instrumentation of the vessel occurs. This phenomenon of lesion architecture changing during revascularization procedures has been encountered by most interventionalists. Plaque redistribution at the carina of the so called “pseudobifurcation” lesion (Lefevre Types 2, 3 and 4) can occur, and resultant reconfigurations may then require the application of bifurcation therapies.

While the Lefevre classification scheme is limited in its prognostication value, it provides a useful nomenclature for describing the following patterns of disease: Type 1: disease involving the main branch (both proximal and distal to the carina) as well as at the side branch ostium; Type 2:disease confined to the proximal and distal main branch, but not involving the side branch; Type 3: disease located only in the main branch proximal to the vessel carina; Type 4: disease confined to the ostium of each branch distal to the carina (4a main branch and 4b side branch) without disease proximal to or at the level of the carina).
While there is no uniform agreement on which of the bifurcation classifications is the most accepted or representative of what is seen in standard practice, the Medina Classification is the most uncomplicated and easiest to remember.²¹ It is a three-digit classification, simply assigning a score of zero or one to each of three regions of the bifurcation lesion dependent on the absence or presence of significant disease. Using the Medina scheme, the ordering of the three numbers are main branch proximal, main branch distal and then side branch (Figure 5). Therefore, a score of 1,0,1 would signify the presence of disease in the main branch proximal, the absence of disease in the main branch distal and the presence of disease in the side branch.
Another anatomic characteristic of lesions having a direct bearing on the outcome of percutaneous revascularization includes the angle at which the side branch arises from its parent vessel.^12,13 There is general acceptance that the so-called Y-shaped lesion (side branch < 70^°) is associated with easier side branch access. Unfortunately, plaque shifting tends to be more pronounced. The T-shaped lesion (side branch angle > 70^°) generally presents greater difficulty with regard to accessing the side branch, but plaque shifting may be less pronounced.
The presence of calcium and fibrous disease plays a major role inprocedure design and outcome. It would seem intuitive that the degree of calcium and fibrous tissue deposition in the vessel as a whole, and specifically at the carina of the bifurcated lesion, should be a predictor of treatment success. Though frequently used in some centers, there is not a well defined role for the atherectomy devices (both rotational and directional) in lesion preparation prior to stenting bifurcations. These modalities have been thought to be useful in well selected cases to achieve side branch preservation when lesions require debulking, as well as preparing heavily calcified vessels for stent implantation. There have been efforts to specifically define a role for lesion debulking of bifurcations in the form of single-center trials. However, the results have generally been inconsistent and these trials may suffer from lack of reproducibility. ^22–24 Our understanding of the effect of calcium on procedural success in bifurcation lesions in the modern era of DES is either mostly anecdotal, based on published observational series, or derived from larger randomized trials not specifically aimed at a bifurcation cohort.^25,26 However, as in all coronary arteries, a vessel laden with a circumferential layer of calcium adjacent to the carina increases the complexity of the lesion and plays a pivotal role in determining procedural success, irrespective of device and/or technique. Under these circumstances, rotational atherectomy or cutting balloon may be a reasonable consideration, either used individually or in combination with each other. Certainly, there are heavily calcified bifurcation lesions in which the only potential opportunity for successful delivery of a stent is dependent on adequate debulking.
When the tissue surrounding the carina of the bifurcation is heavily burdened with fibrous material and calcium, the issue of upfront lesion debulking becomes important, not only for stent delivery to the site of the lesion, but also for optimal stent expansion within the bifurcation. This has been studied extensively in recent years. Relevant studies looking at rotational atherectomy have given variable results; some have shown promise,²⁷ whereas larger-scale randomized trials have failed to support the early enthusiasm for this device synergy.²⁹ The same might be said for the cutting balloon, where small, limited operator, single-center trials gave cause for early optimism.^25–30 In one IVUS-guided cutting balloon plus stenting study in which bifurcations were not excluded,³⁰ the cross-sectional areas measured by IVUS were > 8.0 mm². Again, interinstitutional variability exists and no meaningful generalizations or recommendations can be made on debulking bifurcations.
The bottom line when formulating a treatment strategy: the operator must address the individual morphologic components of the bifurcation in the context of the lesion as a whole. There are a wide variety of anatomic considerations that must be factored in. It is not sufficient to simplify the lesion anatomically based solely on main and side branch distribution of disease; other equally important factors including side branch angulation, extent of lesion calcification and vessel diameter must be considered in devising a rational treatment strategy.
Bifurcation treatment strategies utilizing current DES platforms. Although numerous techniques have been proposed for treating bifurcation CAD, no approach completely circumvents the limitations of the current tubular DES platforms, none of which were designed for the complexities of a branching geometry.^31–40 A single, universally applicable method has not been clearly established. Technical deficiencies include the inability to demonstrate reproducible and durable long-term results, while at the same time gaining acceptance from intermediate skill-level operators based on ease of use. In general, bifurcation methods are broadly classified into 1- or 2-stent approaches, with the two-stent techniques including not only a main branch (MB), but also a side branch (SB) stent.
Main vessel stent only. In contemporary practice, a main vessel (MV) stent-only approach has become the default strategy, although when necessary, this technique can be converted to a provisional T-stent (occasionally culotte or reverse crush as well) when bailout of the side branch is necessary. Studies from the BMS era consistently demonstrated that 1 stent was superior to 2 stents. Al Sawaidi found that major adverse cardiac event (MACE) rates with a single stent was substantially better than more complex 2-stent techniques (26.7% vs. 47.7%).⁴¹ Yamachita reported nearly a 62% restenosis rate when the side and main branches were stented.⁴² More recently, there have been 3 randomized studies with DES comparing simple versus complex stenting strategies. Colombo and colleagues reported their results in 85 patients with 86 bifurcation lesions, all treated with sirolimus-eluting stents (SES).⁴³ The study suffered from a high crossover rate, with nearly half of the single-stent group receiving 2 stents (63 patients receiving MB + SB stents and 22 with MB stenting only). The overall restenosis rate was higher for the 2-stent approach: 28% and 18.7% (p = 0.053). The stent thrombosis rate was 3.5%, all in the MB + SB group. The restenosis rate in the MB was quite low, approximately 5% in both groups. Utilizing the SES as well, Pan and associates randomized 90 patients to either simple stenting (main vessel stent with side branch dilatation) versus complex stenting (main and side branch stenting).⁴⁴ Both strategies were effective and had equivalent clinical outcomes. More recently, Steigen and colleagues from the NORDIC collaborative group, published a large randomized trial comparing MB and MB + SB approaches.⁴⁵ They enrolled 413 patients with true bifurcation lesions and randomized them to either MV stenting or MV + SB stenting (using any 2-stent technique). Unique to this study, 95% of the patients in each randomized group received their assigned therapy; there was little crossover. In the MB group, the primary treatment principles were: 1) stenting of the main vessel; 2) side branch dilatation if there was Thrombolysis In Myocardial Infarction (TIMI) flow of 3 in the side branch; and 3) side branch stenting if TIMI flow was 0 in the side branch after dilatation. In the MB + SB group, the main treatment principles were stenting of both the main vessel and the side branch by application of the crush, culotte or other techniques at the discretion of the operator. In all cases of side branch stenting, theoperator was required to attempt a “kissing balloon” dilatation at the end of the procedure. The side branch length was relatively focal, approximately 6 mm in both groups. There were significantly longer procedural and fluoroscopy times and a larger volume of contrast used in the MB + SB group. Furthermore, an elevation of CK-MB > 3-fold the upper limit of normal was significantly greater in the 2-stent group (18% vs. 8%; p = 0.011). Clinical outcomes were not different (no difference in death, MI, target vessel revascularization or composite MACE). Six-month stent thrombosis was low in both groups (0.5% MB vs. 0% MB + SB). Angiographic follow up at 8 months revealed very low main branch restenosis rates of 4.6% and 5.1% (p = 0.84), and acceptable side branch restenosis rates of 19.2% and 11.5% (p = 0.062) in the MB and MB + SB groups, respectively. These data are compelling, suggesting that a strategy of keeping it simple with 1 main vessel stent is preferred; however, it also suggests that if necessary, a 2-stent approach provides acceptable outcomes.
One practical consideration involves the issue of a final kissing inflation. As a rule, if the side branch is dilated or stented through the struts of the main branch, a final kissing balloon dilatation is required. Dilating through the side of the main branch invariably distorts its architecture (Figure 6), dislocating the side of the stent contralateral to the side branch, sacrificing stent lumen diameter.^46-47 By redilating the MB and SB stents simultaneously, these distortions in the main branch architecture are reversed and lumen diameter is recovered.

Two-stent techniques. In some circumstances, a 2-stent strategy is unavoidable. Examples include when abrupt closure of the side branch occurs and stenting becomes necessary to preserve branch patency and avoid myocardial infarction (MI). In addition, it is very difficult to obtain acceptable results when the side branch plaque is lengthy. Often when the side branch is of significant size (typically > 2.5 mm in diameter) and the side branch plaque length is substantial, an intended/elective 2-stent strategy is best.⁴⁸ Another situation where complex stenting is helpful occurs when the angle of the side branch is such that rewiring is exceptionally difficult. In this situation, using a 2-stent approach that does not sacrifice side branch wire access (at least until the side branch is stented) is optimal. When committed to a 2-stent strategy, which technique is best? There are a myriad of contemporary approaches that can be generally classified as T-stenting with or without crush, V-stenting or simultaneous kissing stents (SKS) and the culotte technique.
T-stenting. T-stenting can be performed in a provisional manner as part of a bailout strategy or can be utilized electively.^49–52 When employed in an elective manner, the classic approach is to place the side branch stent first, taking care to avoid protrusion into the main branch, followed by removal of the side branch wire and stenting of the main branch. The modified T-stenting approach involves aligning the SB and MB stents simultaneously. Following initial deployment of the SB stent, the side branch wire is removed and the main branch is deployed. As in all 2-stent approaches, the side branch must be reaccessed and a final kissing inflation performed. The advantage of this technique is its simplicity, particularly compared to the crush approach, where reaccessing and postdilating the side branch is more difficult; however, because the vast majority of side branches do not arise at right angles, invariably the ostium of the side branch is not fully covered, or excess stent protrudes into the main vessel (Figure 6).
In another variation of this technique, a blocking balloon in the main branch optimizes side branch alignment^53,54 (Figure 7). This self-alignment technique was first described by Rizik and colleagues. In this modified T-stenting approach, wire access is not lost before side branch stenting and branch ostial positioning is simplified.^11,38,53 Following predilatation of the side and main branches, the proximal end of the undeployed branch stent is positioned over the MB wire. The MB balloon and SB stent are then inflated simultaneously. This is a novel modification of the balloon blocking/pullback technique described by Schwartz and Dardas for isolated ostial branch lesions.^54,55 In the 26 patients described, only 15% required further stenting of the ostium at the time of the original procedure, and only 2 patients had angiographic restenosis, both of which were in the side branch ostia, leading to TLR. Although access to the branch is sacrificed while the main branch is stented, the inability to reaccess the stented branch vessel, once treated, is low.
Another variation of the T-stent technique is the TAP (T And small Protrusion) described by Colombo.⁵⁶ This is a variation of the provisional T-stenting where the main vessel stent is implanted, the side branch is reaccessed, redilated, and then a stent is positioned in the side branch with a small amount of stent protruding into the main branch. An MBblocking balloon is inflated simultaneously with SB stent deployment. As always, a final kissing inflation is necessary. This is similar to the self-alignment technique described above, except that it is performed after main branch stenting.
Finally, Rajdev and colleagues have described a “cone crush” modified T-stenting approach.⁵⁷ In this technique, the proximal SB stent is aligned with an uninflated MB-blocking balloon or stent and deployed without inflating the MB device. The SB deployment balloon is retracted several millimeters and inflated to high pressure, creating an ostial flare or cone. The main branch is then stented, covering the cone, and the side branch is reaccessed with a final kissing inflation. Preliminary in vitro and clinical results are encouraging.⁵⁷
An important consideration with the provisional stent technique is whether the jailed side branch lesion is functionally significant. In general, angiography overestimates the severity of these ostial lesions. Koo et al examined 97 ostial side branch lesions with a > 50% diameter stenosis using a pressure wire to determine their functional significance.⁵⁸ No lesions with < 75% diameter stenosis by quantitative coronary angiography had a fractional flow reserve < 0.75 in severity (i.e., none were functionally significant). Of the 73 lesions with > 75% diameter stenosis, only 20 had FFR measurements < 0.75, confirming that even when angiographically severe, jailed ostial side branch stenoses are most often not hemodynamically important.
Standard, modified and reverse crush stenting. Given the often poor coverage of the side branch ostium with standard T-stenting approaches, the crush technique was developed by Colombo to enhance ostial side branch coverage and avoid loss of access to the side branch prior to its being stented. In the standard crush technique, the MB and SB stents are simultaneously positioned such that 2–4 mm of the SB stent protrudes into the main branch (originally more of the SB stent, 5–10 mm, protruded into the main branch).^36,59–62 The MB stent is then positioned and the SB stent is deployed at high pressure. Prior to deploying the MB stent and crushing the side branch, angiography is performed to exclude a distal side branch dissection, a problem easily treated prior to the side branch crush, but sometimes impossible to approach afterwards. Next the SB wire is retracted and the MB stent is inflated at high pressure, crushing the side branch. The side branch is then rewired through the crushed struts and a highpressure > 16 atm dilatation of the side branch is completed with a noncompliant balloon. Often, a hydrophilic wire facilitates side branch rewiring. It is often necessary to predilate with a 1.5 mm rapid-exchange balloon first, as the three layers of stent are often quite difficult to cross with any other balloon. Finally, a high-pressure dilatation in the main branch is performed followed by a kissing inflation, with the balloons sized 1:1 to the two branches. Undersizing the MB balloon during the final kissing inflation leads to distortion of the MB stent, causing suboptimal expansion and apposition of the MB stent.⁶³
Variations of the crush technique include the step crush, in which the SB stent is deployed, followed by crushing with a MB balloon and subsequent stenting of the main branch.³⁷ The later steps are identical to the standard crush. Jim described the “sleeve technique”, a variation of the step crush.⁶² In this approach, following the crush of the side branch with a MB balloon, the side branch is reaccessed and a kissing inflation is performed. The subsequent steps are then identical to the step crush approach. The reverse or internal crush can be used as a provisional or elective technique. The main branch is stented first, followed by reaccess of the side branch. The side branch is then stented, with 2–4 mm of the stent protruding into the main branch. This portion of the SB stent is then crushed internally with a MB balloon. Again, the remainder of the procedure is identical to the standard crush technique, although reaccessing and recrossing the side branch with a balloon is often more difficult.
What have we learned about the crush technique?⁶⁴ Hoye and colleagues describe the results of crush stenting in 231 consecutive patients with 241 bifurcation lesions.⁶⁵ The overall freedom from TLR at 9 months was 90.3%. Left main stenting (approximately 10%) of lesions, was the main predictor of TLR, with an odds ratio of 3.8 compared to non-left main intervention (p < 0.0001). TLR at 9 months was only 5.8% without left main involvement, and was 22.2% with left main involvement. Possible stent thrombosis occurred in 4.3% of patients. Five of the 10 with stentthrombosis had left main stenting. In addition, half of the patients with stent thrombosis had premature cessation of combination antiplatelet therapy. A final kissing inflation had a potent impact on late loss and restenosis. Of the 241 lesions, 124 underwent kissing inflation, kissing inflation being possible in 95% of those in whom it was attempted. A final kissing inflation substantially reduced late loss and restenosis, particularly in the side branch (Figure 8). These results are consistent with Costa’s IVUS observations.⁶⁶ In this report, 25 crush stents underwent IVUS of both branches. IVUS demonstrated incomplete stent apposition in the main vessel segment proximal to the carina in more than 60% of lesions. A minimum lumen area of < 5 mm was found in 76% of the SB stents. These ultrasound studies argue for meticulous dilatation of the side branch with a final kissing inflation.

V-stent or SKS (simultaneous kissing stents) techniques. The simplest of the 2-stent techniques, the V and SKS approaches, consist of positioning 2 stents simultaneously, the proximal portions of the stents forming a double barrel in the main branch and forming a neocarina. The undeployed SB stent is delivered first followed by the MB stent.^67–69 The proximal portions of the stents are aligned, creating a proximal double barrel. When the neocarina (double barrels) extends 5 mm or more into the parent vessel, the technique is considered a SKS approach. The stents should be deployed one at a time (this prevents shifting of the stents, which occurs with simultaneous deployment) at high pressure (at least 16 atm), followed by a simultaneous inflation at 10 atm. Further postdilatation can be completed as necessary, always ending with a simultaneous inflation. Advantages include its simplicity, uninterrupted wire access to both branches, and no need for crossing through the side of a stent for final balloon inflations. This technique is best suited for lesions where the majority of the plaque is distal to the bifurcation, the main branch diameter is larger than both daughter branches, and the main branch plaque is focal. The disadvantage of the technique includes the problem of a proximal edge dissection. If this occurs, placing a proximal bailout stent is difficult, with bias toward one branch inevitable. If a proximal dissection does occur, the double barrels can be extended proximally with 2 more stents, or the V-stent can be converted to a crush by compressing the SB stent with a MB balloon. The side branch is then reaccessed, redilated with a kissing inflation, and finally, a proximal bailout stent is implanted.
Sharma and colleagues reported the outcomes of the SKS technique in 200 consecutive patients.⁷⁰ Procedural success was 100% for the main vessel, and 99% for the side branch; initial clinical success was quite high as well, at 97%. An 8 Fr guide simplifies delivery of the stents. Inhospital and 30-day MACE rates were low, at 3% and 5%, respectively. At a mean follow up of 9 months the TLR rate was only 5%.
In a modified SKS approach, utilized when the proximal disease is lengthy and there is a fear of a dissection, a proximal main vessel stent is deployed, acting as a safety cuff.⁷⁰ Then the side branch is wired, and 2 stents are telescoped through the proximal stent and positioned and deployed in a typical V pattern, their proximal portions lying within the distal aspect of the “cuff” stent.
Culotte technique. The most technically challenging of the two-stent techniques, the culotte approach, has the advantage of optimal coverage of the bifurcation and is suitable for all bifurcation angles.^36,71–74 It is performed by initially stenting the most angulated vessel, typically the side branch. Next, the less angulated vessel is wired, dilated and stented through the struts of the first stent. Finally, the first stent is rewired and final kissing inflation is performed. Like the crush technique, the culotte approach leads to double layers of metal proximally, and rewiring the stents can be very difficult and time consuming. No large series of DES has been published with this technique, although the risks of culotte stenting in the BMS era was associated with poor mid-term results including high target vessel revascularization and stent thrombosis.^41,73