Systematic Review and Network Meta-Analysis Comparing Bifurcation Techniques for Percutaneous Coronary Intervention.

Background Bifurcation lesions account for 20% of all percutaneous coronary interventions and represent a complex subset which are associated with lower procedural success and higher rates of restenosis. The ideal bifurcation technique, however, remains elusive. Methods and Results Extensive search of the literature was performed to pull data from randomized clinical trials that met predetermined inclusion criteria. Conventional meta-analysis produced pooled relative risk (RR) and 95% CI of 2-stent technique versus provisional stent on prespecified outcomes. Both frequentist and Bayesian network meta-analyses were performed to compare bifurcation techniques. A total of 8318 patients were included from 29 randomized clinical trials. Conventional meta-analysis showed no significant differences in all-cause mortality, cardiac death, major adverse cardiac events, myocardial infarction, stent thrombosis, target lesion revascularization, and target vessel revascularization between 2-stent techniques and provisional stenting. Frequentist network meta-analysis revealed that double kissing crush was associated with lower cardiac death (RR, 0.57; 95% CI, 0.38-0.84), major adverse cardiac events (RR, 0.50; 95% CI, 0.39-0.64), myocardial infarction (RR, 0.60; 95% CI, 0.39-0.90), stent thrombosis (RR, 0.50; 95% CI, 0.28-0.88), target lesion revascularization, and target vessel revascularization when compared with provisional stenting. Double kissing crush was also superior to other 2-stent techniques, including T-stent or T and protrusion, dedicated bifurcation stent, and culotte. Conclusions Double kissing crush was associated with lower risk of cardiac death, major adverse cardiac events, myocardial infarction, stent thrombosis, target lesion revascularization, and target vessel revascularization compared with provisional stenting and was superior to other 2-stent techniques. Superiority of 2-stent strategy over provisional stenting was observed in subgroup meta-analysis stratified to side branch lesion length ≥10 mm.

dedicated bifurcation stent

double kissing

final kissing balloon inflation

left main

major adverse cardiac events

proximal optimization technique

side branch

T‐stent or T and protrusion

target lesion revascularization

target vessel revascularization

Clinical Perspective

What Is New?

We used both frequentist and Bayesian approaches of network meta‐analysis in comparing different bifurcation techniques.

We included the findings of newer trials, performed multiple sensitivity analyses, and incorporated results from trials on dedicated bifurcation stents to produce more robust indirect evidence.

What Are the Clinical Implications?

Results of our conventional analysis demonstrated no benefit of 2‐stent strategies over provisional stenting.

Two‐stent strategy should be favored over provisional stenting when lesion length of the side branch is >10 mm.

Double kissing crush technique of bifurcation had more favorable clinical outcomes when compared with provisional stenting, crush, culotte, or T‐stenting or T and protrusion.

Bifurcation lesions account for up to 20% of all percutaneous coronary interventions (PCI) and have been associated with worse clinical outcomes when compared with non‐bifurcation lesions. , Over the years, several bifurcation techniques have been developed to improve procedural and clinical outcomes, but the ideal technique remains elusive. The European Bifurcation Club published its 14th consensus statement in 2019 and advocated for provisional stenting strategy as the standard technique for majority of bifurcation lesions. Upfront 2‐stent approach should be reserved for select cases with appropriate lesion preparation, proximal optimization technique (POT) and final kissing balloon inflation (FKBI). Double kissing (DK) crush received a class IIB recommendation as the choice or upfront 2‐stent technique.

Two previous Bayesian network meta‐analysis have compared the outcomes between different bifurcation techniques but were limited by misclassification and lack of contemporary intervention practices in older trials. , Additional trials comparing bifurcation techniques have since been published, therefore, we performed an updated network meta‐analysis using both frequentist and Bayesian models to compare the various bifurcation techniques.

METHODS

Search Strategy and Inclusion Criteria

The authors declare that all supporting data are available within the article. An extensive literature search was conducted by 2 authors (D.P. and S.A.) using the online libraries, PubMed, Medline, Embase, and Cochrane Library from inception to November 24, 2021. The search terms applied were “bifurcation,” “coronary,” and “randomized trial.” The inclusion criteria were as follows: (1) randomized controlled trials (RCTs) with 1 bifurcation technique in case group and another bifurcation technique in the control group; (2) pre‐specified end points which included all‐cause mortality, cardiac death, major adverse cardiac events (MACE), myocardial infarction (MI), stent thrombosis, target lesion revascularization (TLR), and target vessel revascularization (TVR). Multiple bifurcation techniques could be included in 1 arm if the percentage of each technique was specified. If an RCT had multiple publications, the latest data were collected. For 6 RCTs that included >1 bifurcation technique in 1 arm, , , , , , outcomes were attributed to the predominantly used technique.

Data Extraction and Quality Assessment

Two authors (D.P. and S.A.) collaboratively reviewed full text articles to assess for predetermined eligibility. All the articles were perused for reference citations which were also included if eligible. For each selected RCT, author, published year, follow‐up period, bifurcation techniques, duration of antiplatelet therapy, dual antiplatelet agent, and stent types were arranged into tables (Table 1 and Table S1). Inclusion and exclusion criteria of the RCTs were also summarized in Tables S2 and S3. Anatomical characteristics of bifurcation lesions, demographics, clinical presentation, procedural characteristics, definition of outcomes, and quantitative coronary angiography at baseline were extracted and further organized in Table 2 and Tables S3 through S6. The present meta‐analysis adhered to the Preferred Reporting Items for Systematic Reviews and Meta‐Analyses guidelines. Risk of biases were assessed using Cochrane Collaboration’s tool (Table S7). Only data from published papers that are publicly available were used, so the study was not under the purview of the institutional review board.

Table 1

Characteristics of Selected Trials

Trial	Author	Y	Follow‐up	DAPT	Left main	Total	Predominant bifurcation		Control
							Technique	n	Technique	n
EBC MAIN 9	Hildick‐Smith et al	2021	1 y	6 mo	Yes	467	Culotte (53%), T/TAP (32%), DK crush (5%), missing (7%)	237	PS	230
NBBS IV 10	Kumsars et al	2020	2 y	12 mo	Yes	446	Culotte (66%), crush (22%), T‐stent (7%), others (6%)	228	PS	218
DEFINITION II 13	Zhang et al	2020	1 y	12 mo	Yes	653	DK crush (78%), culotte (18%), TAP (3%), others (1%)	328	PS	325
DKCRUSH‐V 16	Chen et al	2019	3 y	12 mo	Yes	482	DK crush	240	PS	242
COBRA 17	Bennett et al	2018	5 y	12 mo	No	40	DBS	20	Culotte	20
DKCRUSH‐II 18	Chen et al	2017	5 y	12 mo	Yes	366	DK crush	183	PS	183
BBK II 19	Ferenc et al	2016	1 y	6 mo	Yes	300	Culotte	150	TAP	150
POLBOS II 20	Gil et al	2016	1 y	12 mo	Yes	202	DBS	102	PS	100
EBC TWO 21	Hildick‐Smith et al	2016	1 y	12 mo	No	200	Culotte	97	PS	103
SMART‐STRATEGY 22	Song et al	2016	3 y		Yes	258	TAP	130	PS	128
Zhang et al (2016) 23	Zhang et al	2016	9 mo	12 mo	Yes	104	Culotte	52	PS	52
Zheng et al (2016) 24	Zheng et al	2016	1 y	12 mo	Yes	300	Crush	150	Culotte	150
DKCRUSH‐III 25	Chen et al	2015	3 y	12 mo	Yes	415	DK crush	208	Culotte	207
BBK I 26	Ferenc et al	2015	5 y	6 mo	No	202	T‐stent	101	PS	101
TRYTON 27	Genereux et al	2015	9 mo	6–12 mo	No	704	DBS	355	PS	349
POLBOS I 28	Gil et al	2015	1 y	12 mo	Yes	243	DBS	120	PS	123
PERFECT 29	Kim et al	2015	1 y	12 mo	No	419	Crush	213	PS	206
NSTS 30	Kervinen et al	2013	3 y	6–12 mo	Yes	424	Crush	209	Culotte	215
NBS 12	Maeng et al	2013	5 y	6–12 mo	Yes	404	Crush (50%), culotte (21%), others (29%)	202	PS	202
Ruiz‐Salmeron et al (2013) 31	Ruiz‐Salmeron et al	2013	9 mo	12 mo	No	65	T‐stent	34	PS	31
Ye et al (2012) 32	Ye et al	2012	1 y	12 mo	No	68	DK crush	38	PS	30
BBC ONE 8	Hildick‐Smith et al	2010	9 mo	9 mo	No	500	Crush (68.1%), culotte (30.2%), others (1.6%)	250	PS	250
Lin et al (2010) 11	Lin et al	2010	8 mo	12 mo	No	108	DK crush (65%), culotte (25%), others (10%)	54	PS	54
Ye et al (2010) 33	Ye et al	2010	8 mo	12 mo	No	51	DK crush	25	PS	26
CACTUS 34	Colombo et al	2009	6 mo	6 mo	No	350	Crush	177	PS	173
Cervinka et al (2008) 35	Cervinka et al	2008	1 y	1 mo	No	60	DBS	30	PS	30
DKCRUSH‐I 36	Chen et al	2008	8 mo	12 mo	Yes	311	DK crush	155	Crush	156
Colombo et al (2004) 37	Colombo et al	2004	6 mo	3 mo	No	85	T‐stent	63	PS	22
Pan et al (2004) 38	Pan et al	2004	6 mo	12 mo	Yes	91	T‐stent	44	PS	47

BBC ONE indicates British Bifurcation Coronary Study; BBK I, Bifurcations Bad Krozingen I; BBK II, Bifurcations Bad Krozingen II; CACTUS, Coronary Bifurcations: Application of the Crushing Technique Using Sirolimus‐Eluting Stents; COBRA, Complex Coronary Bifurcation Lesions: Randomized Comparison of a Strategy Using a Dedicated Self‐Expanding Biolimus‐Eluting Stent Versus a Culotte Strategy Using Everolimus‐Eluting Stents; DBS, dedicated bifurcation stent; DEFINITION II, Definitions and Impact of Complex Bifurcation Lesions on Clinical Outcomes After Percutaneous Coronary Intervention Using Drug‐Eluting Stents; DKCRUSH‐I, Study Comparing the Double Kissing Crush With Classical Crush for the Treatment of Coronary Bifurcation Lesions; DKCRUSH‐II, Double Kissing Crush Versus Provisional Stenting Technique for Treatment of Coronary Bifurcation Lesions; DKCRUSH‐III, Double Kissing Crush Versus Culotte Stenting for the Treatment of Unprotected Distal Left Main Bifurcation Lesions; DKCRUSH‐V, Double Kissing Crush Versus Provisional Stenting for Left Main Distal Bifurcation Lesions; DK crush, double kissing crush; EBC MAIN, European Bifurcation Club Left Main Coronary Stent Study; EBC TWO, European Bifurcation Coronary TWO; NBBS IV, Nordic‐Baltic Bifurcation Study IV; NBS, Nordic Bifurcation Study; NSTS, Nordic Stent Technique Study; PERFECT, Optimal Stenting Strategy for True Bifurcation Lesions; POLBOS I, PoLish Bifurcation Optimal Stenting I; POLBOS II, Polish Bifurcation Optimal Stenting II; PS, provisional stent; SMART‐STRATEGY, Smart Angioplasty Research Team‐Optimal STRATEGY for Provisional Side Branch Intervention in Coronary Bifurcation Lesions; T/TAP, T‐stenting or T and protrusion; and TRYTON, Prospective, Single Blind, Randomized Controlled Study to Evaluate the Safety & Effectiveness of the Tryton Side Branch Stent Used With DES in Treatment of de Novo Bifurcation Lesions in the Main Branch & Side Branch in Native Coronaries.

Table 2

Demographics, Clinical Presentation, and Characteristics of Lesion

Case/control, %	EBC MAIN 9	NBBS IV 10	DEFINITION II 13	DKCRUSH‐V 16	COBRA 17	DKCRUSH‐II 18	BBK II 19	POLBOS II 20	EBC TWO 21
Age, y	71.4/70.8	63.0/64.0	63.0/64.0	65.0/64.0	66.0/64.0	63.9/64.7	66.3/69.1	67.2/66.6	63.5/62.9
Male	74/79	…	77.7/76.9	82.9/77.7	14.0/15.0	78.8/75.8	71.3/76.0	76.9/75.0	78.0/85.0
BMI, mean	28.4/28.6	…	24.8/24.7	…	…	…	…	…	…
Diabetes	27.0/29.0	15.4/16.5	34.1/35.7	28.8/25.6	25.0/20.0	19.6/23.1	27.3/28.0	44.1/32.0	31.0/25.0
Hypertension	82.0/79.0	65.6/70.0	66.2/70.1	72.9/64.5	75.0/70.0	65.2/60.9	88.0/85.3	84.3/81.0	68.0/63.0
Dyslipidemia	72.0/70.0	81.1/82.0	69.2/68.6	47.5/47.5	95.0/95.0	33.7/29.1	…	83.3/81.0	70.0/70.0
Smoking	13.0/16.0	21.1/18.9	28.4/30.2	…	25.0/20.0	…	11.3/11.3	20.6/26.0	50.0/56.0
PVD	16.0/14.0	…	5.8/4.6	…	…	…	…	3.9/9.0	8.0/6.0
Renal failure	4.0/5.0	…	…	…	…	…	…	10.8/7.0	…
Family history	33.0/33.0	47.4/50.0	…	…	…	…	40.7/39.3	…	…
Previous MI	28.0/26.0	…	11.9/12.9	21.7/21.1	30.0/10.0	17.4/14.2	16.0/21.3	43.1/48.0	41.0/39.0
Previous PCI	43.0/41.0	33.5/35.5	19.8/16.6	…	40.0/20.0	21.2/20.9	38.0/32.0	52.0/57.0	41.0/40.0
Previous stroke	7.0/7.0	…	…	…	…	…	…	…	…
LVEF, mean	…	56.0/57.0	59.0/60.0	…	67.0/68.0	…	56.0/57.0	…	…
Stable CAD	60.0/66.0	…	…	…	80.0/80.0	…	…	…	68.0/69.0
Stable angina	…	82.4/86.6	24.1/21.8	…	…	15.3/11.0	…	…	…
Silent ischemia	…	1.3/0.5	5.2/5.2	…	…	1.6/3.8	…	…	…
ACS	40.0/33.0	…	…	…	…	…	21.3/19.3	…	32.0/31.0
Unstable angina	…	16.7/12.9	48.8/50.5	…	20.0/20.0	66.8/68.7	…	…	…
Acute MI	…	…	22.0/22.5	…	…	16.3/16.3	…	…	…
SYNTAX, mean	23.2/22.6	…	24.7/24.2	…	…	…	…	17.5/18.2	…
0–22	26.0/30.0	…	44.8/48.6	…	…	…	…	…	…
22–32	57.0/56.0	…	33.8/32.6	…	…	…	…	…	…
>32	…	…	21.3/18.8	37.9/36.4	…	…	…	…	…
Medina class
1,0,0	0/0	…	…	…	…	…	…	…	…
0,1,0	0/0	…	…	…	…	…	…	…	…
1,1,0	0/0	…	…	…	0/5.0	…	…	…	…
1,1,1	89.0/90.0	…	86.3/82.5	…	50.0/70.0	84.2/78.7	…	…	68.0/81.0
0,0,1	0/0	…	…	…	…	…	…	…	…
1,0,1	0/0	…	…	…	15.0/10.0	…	…	…	7.0/6.0
0,1,1	11.0/10.0	…	12.5/14.5	…	35.0/15.0	15.8/21.3	…	…	24.0/12.0
Complex features
Trifurcation	4.0/5.0	…	9.5/6.8	…	…	…	…	…	…
Calcification	54.0/44.0	43.6/48.4	38.7/40.3	37.1/39.7	…	…	…	…	17.0/19.0
Tortuosity	24.0/19.0	7.0/2.8	…	…	…	…	…	…	15.0/10.0
Lesion location
Left main	100/100	1.3/2.77	28.7/28.9	100/100	…	17.8/15.7	18.7/15.3	35.3/38	…
LAD	…	76.7/74.2	62.5/60.6	…	≥95/≥95	60.5/59.5	54.7/55.3	44.1/43	77/78
LCx	…	17.6/16.6	5.2/7.7	…	…	12.4/16.2	24.0/25.3	15.7/15.0	19/15
RCA	…	4.0/6.5	3.7/2.8	…	…	9.2/8.6	2.7/4.0	4.9/4.0	4/6

ACS indicates acute coronary syndrome; BBC ONE, British Bifurcation Coronary Study; BBK I, Bifurcations Bad Krozingen I; BBK II, Bifurcations Bad Krozingen II; CACTUS, Coronary Bifurcations: Application of the Crushing Technique Using Sirolimus‐eluting stents; CAD, coronary artery disease; COBRA, Complex Coronary Bifurcation Lesions: Randomized Comparison of a Strategy Using a Dedicated Self‐Expanding Biolimus‐Eluting Stent Versus a Culotte Strategy Using Everolimus‐Eluting Stents; DEFINITION II, Definitions and Impact of Complex Bifurcation LesIons on Clinical Outcomes After Percutaneous Coronary Intervention Using Drug‐Eluting Stents; DKCRUSH‐I, Study Comparing the Double Kissing Crush with Classical Crush for the Treatment of Coronary Bifurcation Lesions; DKCRUSH‐II, Double Kissing Crush Versus Provisional Stenting Technique for Treatment of Coronary Bifurcation Lesions; DKCRUSH‐III, Double Kissing Crush Versus Culotte Stenting for the Treatment of Unprotected Distal Left Main Bifurcation Lesions; DKCRUSH‐V, Double Kissing Crush Versus Provisional Stenting for Left Main Distal Bifurcation Lesions; EBC MAIN, European Bifurcation Club Left Main Coronary Stent Study; EBC TWO, European Bifurcation Coronary TWO; LAD, left anterior descending coronary artery; LCx, left circumflex coronary artery; MI, myocardial infarction; NBBS IV, Nordic‐Baltic Bifurcation Study IV; NBS, Nordic Bifurcation Study; NSTS, Nordic Stent Technique Study; PERFECT, Optimal Stenting Strategy for True Bifurcation Lesions; POLBOS I, Polish Bifurcation Optimal Stenting I; POLBOS II, Polish Bifurcation Optimal Stenting II; SMART‐STRATEGY, Smart Angioplasty Research Team‐Optimal STRATEGY for Provisional Side Branch Intervention in Coronary Bifurcation Lesions; PVD, peripheral vascular disease; RCA, right coronary artery; and TRYTON, Prospective, Single Blind, Randomized Controlled Study to Evaluate the Safety & Effectiveness of the Tryton Side Branch Stent Used With DES in Treatment of de Novo Bifurcation Lesions in the Main Branch & Side Branch in Native Coronaries.

Statistical Analysis

For conventional meta‐analysis, random effects model based on DerSimonian and Laird method was used to produce pooled relative risk (RR) and 95% CI of 2‐stent technique versus provisional stent on prespecified outcomes. Haldane‐Ascombe corrections were made for zero‐cell corrections. Egger and Begg‐Mazumdar tests were applied after visualization of funnel plots to evaluate for publication biases (Table S8). Both Cochran’s Q and Higgins and Thompson’s I2 statistic were generated to describe the heterogeneities among the trials. P value <0.05 or 95% CI not including 1 was statistically significant.

Network meta‐analysis based on frequentist framework was first performed to produce network estimates from direct and indirect estimates. To evaluate for inconsistencies, node‐splitting analysis was conducted to compare direct and indirect evidence for each outcome. P value <0.05 signified the presence of inconsistency. Tau‐squared and I2 were used to assess the heterogeneities in the network models, which was then broken down into heterogeneities within designs and between designs, each evaluated with Cochran’s Q (Table S9). P‐scores of each bifurcation technique were also calculated for all outcomes (Table S10). P‐scores were interpreted only for outcomes in which the network meta‐analysis showed significant difference among the bifurcation techniques. Bayesian network meta‐analysis was additionally performed whereby estimates of the bifurcation techniques were calculated through a generalized linear model fitted under a hierarchic Bayesian random‐effect framework. Models were computed by Markov‐chain Monte Carlo simulations using 4 chains, 5000 adaptations, and 100 000 iterations. Convergence was observed by visual inspection of time‐series and density plots. Surface under the cumulative ranking scores were calculated from the Bayesian model to validate the P‐scores from the frequentist model (Table S11). Hierarchy of bifurcation techniques were then displayed using rankograms (Table S12). Frequentist network meta‐analysis was performed with meta and netmeta packages, and Bayesian network meta‐analysis with gemtc and rjags packages, all with the use of R version 4.0.5 (R Foundation for Statistical Computing, Vienna, Austria).

RESULTS

Bibliographic Search and Trial Characteristics

After a comprehensive search of the literature, 29 RCTs, published from 2004 to 2021, were included in the study (Figure 1). A total of 8318 patients were included consisting of 3225 provisional stenting, 1357 crush, 1356 culotte, 1231 DK crush, 627 dedicated bifurcation stent (DBS), and 522 T‐stent or T and protrusion (T/TAP) (Table 1). The follow‐up period ranged from 6 months to 5 years. Left main (LM) bifurcations were included in 16 trials† while solely non‐LM bifurcations were included in 13 trials.‡ Most of the trials prescribed clopidogrel as the dual antiplatelet agent, with some older studies also administering ticlopidine (Table S1). The types of stents used varied widely across the studies. Demographics, clinical presentation, and characteristics of bifurcation lesions in each of the trials were also heterogeneous (Table 2). Clinical and anatomical inclusion criteria are summarized in Tables S2 and S3. Details of quantitative angiography and PCI procedural information were inconsistent across all trials (Tables S4 and S6). Angiographic follow‐ups were provided for most trials and ranged between 6 and 13 months after index procedure (Table S13).

Figure 1

Flow diagram of the search for relevant trials.

The flow diagram illustrates the process of searching and screening the databases to identify trials that meet the prespecified inclusion criteria.

Comparison of Bifurcation Techniques

Conventional meta‐analysis was initially performed to compare clinical outcomes between 2‐stent and provisional stent strategies. There were no significant differences in all‐cause mortality, cardiac death, MACE, MI, stent thrombosis, TLR, and TVR (Figures S1 through S7). However, in subgroup analysis stratified to the length of side branch (SB) lesion, 2‐stent strategies performed better than provisional stents at reducing cardiac death (RR, 0.60; 95% CI, 0.40–0.90), MACE (RR, 0.68; 95% CI, 0.50–0.93), TLR (RR, 0.55; 95% CI, 0.39–0.78), and TVR (RR, 0.58; 95% CI, 0.36–0.95) when the lesion in the SB was ≥10 mm (Figures S8 through S14). On the other hand, the risk of MACE (RR, 1.20; 95% CI, 1.00–1.44) was marginally greater in 2‐stent strategy than provisional stenting when the length of the SB lesion was <10 mm (Figure S10).

Frequentist network meta‐analysis (Figure 2) revealed that DK crush was associated with lower cardiac death (RR, 0.57; 95% CI, 0.38–0.84), MACE (RR, 0.50; 95% CI, 0.39–0.64), MI (RR, 0.60; 95% CI, 0.39–0.90), stent thrombosis (RR, 0.50; 95% CI, 0.28–0.88), TLR (RR, 0.44; 95% CI, 0.33–0.59), and TVR (RR, 0.48; 95% CI, 0.34–0.66) when compared with provisional stenting (Figure 3). T/TAP performed worse than provisional stenting and was associated with increased risk of stent thrombosis (RR, 2.37; 95% CI, 1.02–5.51). Node‐splitting analysis showed no inconsistencies between direct and indirect evidence for all outcomes (Table S14).

Figure 2

Network plot of selected trials.

The network plot demonstrates the number of studies and patients included among trials that compared double kissing crush, dedicated bifurcation stent, culotte, crush, provisional stenting, and T‐stent or T and protrusion. The size of the blue circles and blue lines are proportional to the total sample size and number of relevant studies, respectively. DBS indicates dedicated bifurcation stent; DK, double kissing; PS, provisional stenting; and T/TAP, T‐stent or T and protrusion.

Figure 3

Network meta‐analysis of bifurcation techniques with provisional stenting as reference.

The figures show the relative risk of each bifurcation technique compared to provisional stenting for 7 different outcomes. The vertical line inside the blue box represents the relative risk and the perpendicular horizontal line represents the 95% CI. Relative risk above 1 favors provisional stenting (red arrow) whereas that below 1 favors the compared bifurcation technique (blue arrow). DBS indicates dedicated bifurcation stent; DK, double kissing; RR, relative risk; and T/TAP, T‐stent or T and protrusion.

DK crush was associated with lower risk of cardiac death, MACE, MI, stent thrombosis, TLR, and TVR compared with crush. It was also associated with lower risk of MACE, MI, stent thrombosis, TLR, and TVR compared with culotte. Similarly, DK crush was associated with better outcomes compared with DBS, T/TAP, and provisional stent (Table S15). The superiority of DK crush was consistently observed in sensitivity analysis of trials that only included true bifurcations (Table S16) or those that excluded LM bifurcations (Table S17). Similar outcomes were found on sensitivity analysis excluding trials without LM bifurcations (Table S18). After excluding trials allowing multiple bifurcation techniques in 1 arm, outcomes still favored DK crush (Table S19).

P‐scores calculated by the frequentist model demonstrated that DK crush ranked the highest for MACE, MI, stent thrombosis, TLR, and TVR, most often followed by provisional strategy (Figure 4). Culotte, crush, and T/TAP were associated with lower ranks. Surface under the cumulative ranking scores from Bayesian model produced identical results (Figures S15 and S16). Rankograms redemonstrated the superiority of DK crush and the inferiority of culotte, crush, and T/TAP (Figure 5).

Figure 4

Bar graph showing P‐scores of each bifurcation technique for every outcome.

The bar graphs show the P‐scores of provisional stenting (gray), culotte (lavender), DK crush (blue), dedicated bifurcation stent (green), T‐stent or T and protrusion (orange), and crush (red) from the frequentist network meta‐analysis for each outcome. P‐scores measure the extent of certainty that the bifurcation technique is better than competing techniques. DBS indicates dedicated bifurcation stent; DK, double kissing; and T/TAP, T‐stent or T and protrusion.

Figure 5

Rank probability analysis for outcomes of interest.

Displayed as rankograms, results of rank probability analysis show the probability of culotte (lavendar), double kissing crush (blue), dedicated bifurcation stent (green), crush (red), T‐stent or T and protrusion (orange), and provisional stenting (gray) being the best, second, third, fourth, fifth, and sixth for each of the outcomes. The x‐axis and y‐axis represent the rank and probability, respectively. DBS indicates dedicated bifurcation stent; DK, double kissing; PS, provisional stenting; and T/TAP, T‐stent or T and protrusion.

DISCUSSION

The results from our present comprehensive meta‐analysis show that DK crush was superior to other bifurcation techniques in reducing the risk of not only stent thrombosis, TLR, TVR, but also the hard end points including cardiac death, MACE, and MI. Subgroup analysis within conventional meta‐analysis demonstrated upfront 2‐stent strategy was superior to provisional stenting when the SB lesion length was ≥10 mm.

Two previous network meta‐analyses have been reported comparing different bifurcation techniques. , Crimi et al compiled 26 RCTs and showed that DK crush was associated with the lowest device‐oriented clinical event consisting of cardiac death, target‐vessel MI, stent thrombosis, TLR, and TVR. This was in line with the findings of the present network meta‐analysis, which also presented highest P‐scores as well as surface under the cumulative ranking scores for DK crush. However, Crimi et al mislabeled THUEBIS (Thueringer bifurcation Study) trial as a trial on DBS versus provisional stent and did not account for case groups in which >1 bifurcation techniques were used. Di Gioia et al performed a similar meticulous network meta‐analysis on the same subject that included 21 RCTs and 3 sensitivity analyses. They showed that DK crush was associated with lower MACE which was driven by lower rates of TLR and TVR. However, they did not include trials with DBS, and their sensitivity analysis on trials with only non‐LM bifurcations included a significant number of trials with LM bifurcations.

Several theories have been proposed to explain the superiority of DK crush over more conventional and less complex bifurcation techniques. DK crush is advantageous in that it is not affected by the bifurcation angle and maintains wire access in the MV. , However, it is a complex multi‐step process that requires crossing stent struts twice and results in greater radiation and contrast exposures. FKBI, an important step in the DK crush technique, may also be contributing to favorable outcomes. Summary of procedural characteristics in our analysis also showed that FKBI was performed more frequently in the DK crush arms (Table S4). Bench modeling demonstrated greater occurrence of stent malapposition in single kissing compared with DK with FKBI. Chen et al claimed that DK crush reduced the strut layer in the SB ostium, thereby increasing the success of the final kissing balloon inflation. Ye et al also explained the superiority of DK crush with the higher rate of FKBI, which potentially leads to improved stent apposition, optimized stent geometry, and reduced flow disturbance. , However, FKBI after PCI of distal LM bifurcation lesions was not associated with improved outcomes within the EXCEL (Evaluation of XIENCE Versus Coronary Artery Bypass Surgery for Effectiveness of Left Main Revascularization) trial. Similarly, in DEFINITION II (Definitions and Impact of Complex Bifurcation LesIons on Clinical Outcomes After Percutaneous Coronary Intervention Using Drug‐Eluting Stents) trial, despite having similar percentages of FKBIs in both arms, outcomes still favored the 2‐stent strategy (78% of which was DK crush).

Dedicated bifurcation stents, despite their initial promise to address the limitations of standard 2 stent strategies, did not show any benefit over conventional stenting techniques. Although the results of our analysis gave high ranking to DBS for all‐cause mortality and cardiac death, these results are not statistically significant and hence this finding is unfounded (Table S15). The accuracy of P‐scores and surface under the cumulative ranking scores are compromised with non‐significant results and the higher ranks of DBS likely occurred because of the relatively smaller sample and low number of events.

Crush technique has been a historic favorite and has continued to evolve over time with smaller and smaller protrusion of the side branch stent: classic, mini, and nano. The included trials have predominantly used the classic crush technique, except for the PERFECT (Optimal Stenting Strategy for True Bifurcation Lesions) trial, which used the mini‐crush technique. Results could be anticipated to be different if more trials had used mini‐crush or nano‐crush techniques and will need future trials to evaluate further.

Coronary intravascular imaging has shifted the paradigm of coronary interventions from “guessing” based on quantitative angiography to “knowing” accurate vessel and lesion characteristics. Recent meta‐analysis demonstrated that the use of intravascular ultrasound during bifurcation PCI was associated with lower risk of MACE compared with angiography‐guided PCI. Only 9 out of the 29 RCTs included in our present study reported the use of intravascular ultrasound,§ which was not significantly different between the 2 arms except in the PERFECT trial (Table S4). Similarly, only European Bifurcation Club MAIN reported the percentage of optical coherence tomography in both its arms. Perhaps a more standardized and uniform use and reporting of imaging would improve our understanding of bifurcation lesion preparation and optimization.

Proximal optimization technique, which was introduced in 2010, has been proposed to reconstruct the natural and fractal geometry of coronary bifurcations and achieve optimal stent expansion and apposition in the proximal segment. POT was strongly recommended by European Bifurcation Club across all 2 stent strategies. However, it was reported in only 5 RCTs. , , , , The difference in POT percentage was not significant between the 2 arms in the European Bifurcation Club MAIN (European Bifurcation Club Left Main Coronary Stent Study [EBC MAIN]) and DKCRUSH‐V (Double Kissing Crush versus Provisional Stenting for Left Main Distal Bifurcation Lesions) trials, but was significantly different in DEFINITION II and POLBOS I and II (Polish Bifurcation Optimal Stenting Polish Bifurcation Optimal Stenting) trials (Table S4). Analysis of e‐ULTIMASTER (Prospective, Single‐Arm, Multi Centre Observations Ultimaster Des) multinational registry showed that POT was associated with reduction in target lesion failure and stent thrombosis regardless of bifurcation anatomy and technique, so the potential differences in the use of POT could have also affected the studied outcomes.

Other potential confounding variables such as the site of access (radial versus femoral) were only reported in 10 RCTs. , , , , , , , , , However, the difference was not significant between any of the 2 arms (Table S4), and a propensity‐matched analysis of Coronary Bifurcation Stenting Registry in Korea also did not observe any differences in cardiac death, MI, TLR, and MACE between transradial and transfemoral approaches. Antiplatelet use after bifurcation stenting varied across trials. More recent trials , included ticagrelor and prasugrel as their choice of antiplatelet, while most other studies used clopidogrel and some older studies , , , used ticlopidine (Table S1). A recent network meta‐analysis suggested that ticagrelor and prasugrel performed better than clopidogrel in reducing cardiovascular outcomes, so choice of dual antiplatelet agent also need dedicated analysis in bifurcation lesions. Furthermore, the heterogeneities in the anatomical characteristics of the bifurcation lesions in each of the RCTs were greatly variable, with some trials including more complex cases as those described in the DEFINITION criteria, limiting our results to be applied across all patient population (Tables S4 and S7).

Limitations

The present network meta‐analysis is subject to several limitations. Individual patient level data were unavailable. The time RCTs were conducted spanned from the year 2004 to 2021, and there have been significant improvements in secondary prevention, stent design, choice of anti‐proliferative agent in stents, and functional testing of lesions. Events were attributed to the most performed bifurcation technique in 6 RCTs , , , , , where >1 technique was performed. Sensitivity analyses accounting for heterogeneities among the trials were conducted, which yielded similar results (Tables S15 through S19). Several RCTs were subject to high risk of bias primarily attributable to the lack of blinding and conducting open‐label studies. Operators could not be masked because of the nature of the study. Significant crossovers occurred in many of the provisional stent arms, and many of the RCTs were conducted by the same group of experts at high‐volume centers who would be more proficient in performing complex interventions. Outcomes after DK crush may vary depending on the level of expertise of operators, so further studies reproducing similar safety and efficacy will be required to validate the superiority of the DK crush technique.

CONCLUSIONS

The findings of the present network meta‐analysis of bifurcation techniques showed that DK crush was associated with lower risk of cardiac death, MACE, MI, stent thrombosis, TLR, and TVR compared with provisional stenting. Superiority of 2‐stent strategy over provisional stenting was observed in subgroup meta‐analysis stratified to SB lesion length ≥10 mm. Given the findings from successive network meta‐analysis, including our present study, and those from DEFINITION II and DKCRUSH‐V trials, DK crush can be considered over other 2‐stent strategies in patients with complex bifurcation lesions. Further studies will be required to reproduce and validate these findings.

Sources of Funding

None.

Disclosures

None.

Tables S1–S19

Figures S1–S16

Click here for additional data file.