This is a single-center randomized pilot study. Patients with the following criteria were deemed eligible: (1) de novo true BLs (Medina type 1, 1, 1; 0, 1, 1; 1, 0, 1) and (2) SB $\ge$2.25 mm by visual estimation; Patients with the following criteria were excluded: (1) left main BLs, (2) other lesions requiring PCI in addition to the target BLs, and (3) lesions unsuitable for DCB treatment because of bifurcation or anatomy features (e.g., severe calcification, tortuous lesions, and wiring difficulty, etc.), (4) ST-elevation myocardial infarction (MI) within 48 h, (5) high bleeding risks, (6) allergy to any drugs needed, and (7) life expectancy 1 year.

From Feb. 2019 to Feb. 2021, a total of 60 patients were randomized at a 1:1 ratio to receive either a DCB-based strategy or a DST-based strategy for BL intervention and then scheduled follow-up (Fig. 1). The protocol was approved by the Ethics Committee of Fujian Medical University Union Hospital (Supplementary Approval File No 2019KY035). All patients gave written informed consent.

Fig. 1.

Study Flowchart. *, Bailout stenting of MV only or both MV and SB was allowable in lesion pretreatment or in DCB treatment if there were unacceptable results. ${}^{\mathrm{\#}}$, DST with using DK-crush, DK-culotte or T-stent was left at discretion of the operators. DCB, drug-coated balloon; DST, dual-stenting techniques; MV, main-vessel; SB, side-branch.

DCB-based strategy: This technique is a combined approach, characterized by DCB-centered angioplasty, optimal lesion pretreatment, and allowable use of bailout stenting and GP IIb/IIIa inhibitors, to ensure procedural safety. The key steps (Fig. 2) are briefly described below: (1) Scoring or cutting balloons was preferred for lesion preparation, and pre-dilating with smaller plain balloons for subsequent passage of scoring or cutting balloons or post-dilating with larger non-compliant balloons for achievement of an optimal lumen was allowable. (2) After optimal lesion preparation of the MV and SB, DCB angioplasty was performed on the SB and then the MV, and final kissing dilation was at the discretion of the operators. (3) The diameter of proximal and distal MB was averaged as the reference vessel diameter (RVD) of MV, a balloon to RVD ratio of $\approx$1.0 was adopted in the final lesion preparation and DCB angioplasty. (4) Bailout stenting for MV or MV+SB was allowable if unacceptable results [14, 15, 16] were obtained in lesion preparation or DCB angioplasty stage.

Fig. 2.

The Procedural Steps of DCB-based Strategy. *, Unacceptable results were defined as any of residual stenosis $>$30% or flow-limiting dissection either in MV, SB or both. ${}^{\mathrm{\#}}$, Bailout stenting of MV only or both MV and SB was left at discretion of the operators. DCB, drug-coated balloon; MV, main-vessel; SB, side-branch.

DST-based strategy: One of the DSTs (DK-crush, DK-culotte or T-stenting) may be selected and should be completed according to the standards of various DSTs [2, 3].

The DCB was a paclitaxel/iopromide matrix coating balloon (SeQuent® Please, B. Braun Melsungen AG, Germany). All stents were the 2nd generation drug-eluting stents, including Resolute${}^{\text{TM}}$ (Medtronic, Minneapolis, MN, USA), Xience${}^{\text{TM}}$ (Abbott Vascular, Santa Clara, CA, USA), Firebird-2${}^{\text{TM}}$ (Microport, Shanghai, China), and Excel${}^{\text{TM}}$ (JW, Shandong, China).

All patients received pretreatment with aspirin and P2Y12 antagonists of clopidogrel or ticagrelor with a loading dose as indicated. Intra-procedural heparin (70–100 U/kg) was intravenously injected with a supplemented bolus of 1000 U given per hour to maintain an activated clotting time of 250–300 seconds. Peri-procedural use of glycoprotein IIb/IIIa inhibitors was allowable at the operator’s discretion. Dual anti-platelet therapy with aspirin plus clopidogrel or ticagrelor (preferred) was maintained for one year for both strategies, followed by indefinite single anti-platelet therapy (aspirin, clopidogrel or ticagrelor).

Coronary angiography (CAG) was performed pre-procedurally, post-procedurally, and at follow-up after intracoronary injection of 200 $\mu$g nitroglycerin.

For quantitative coronary angiographic analysis (QCA), the bifurcation was simply segmented into: (1) MV, the segment from the proximal to distal end treated by stents or DCBs; and (2) SB, the segment from the carina to distal end treated by stents or DCBs. The reference vessel diameter (RVD) of the MV was the averaged diameter of the proximal and distal MB, and the minimal lumen diameter (MLD) was directly measured at the narrowest site. The diameter stenosis percent (DSP) was calculated by (RVD-MLD) / RVD $\times$ 100%, and the late lumen loss (LLL) was calculated as the post-procedural MLD — follow-up MLD. Binary restenosis (BRS) was defined as DSP $>$50%.

Clinical data were collected during the hospital stay and by hospital visit or telephone contact at 1, 3, 6, 9, and 12 months after discharge and afterward annually thereafter. Follow-up CAG was scheduled at 12 $\pm$ 1 months post-procedurally.

All deaths were deemed cardiogenic unless there was clear evidence of non-cardiac causes. Peri-procedural MI (within 48 h) was defined as: I. a creatine kinase-MB (CK-MB) $>$10 or troponin $>$70 $\times$ the upper reference limit (URL), or II. a CK-MB $>$5 or troponin $>$35 $\times$ URL plus either: (1) new pathological Q waves in $\ge$2 contiguous leads or new left bundle branch block; (2) angiographically documented graft or coronary artery occlusion or new severe stenosis with thrombosis; (3) imaging evidence of new loss of viable myocardium; or (4) new regional wall motion abnormality. In non-ST elevation MI (NSTEMI) patients with elevated pre-procedural biomarkers in whom the levels were stable ($\le$20% variation) or falling, peri-procedural MI could be diagnosed when the post-procedural biomarkers rise by $>$20% along with the criteria similar to the aforementioned Definition II. Spontaneous MI (after 48 h) was defined as a clinical syndrome consistent with MI along with a CK-MB or troponin $>$1 $\times$ URL and new ST-segment elevation or depression or other findings as described above. All MI were considered target vessel myocardial infarction (TVMI) unless there was clear evidence attributable to a non-target vessel. Clinically driven TLR/TVR was defined as typical angina pectoris or confirmed ischemia referable to the target lesion/vessel requiring urgent or selective repeat PCI or coronary artery bypass graft. TVST was determined according to the ARC classification [17]. The major cardiac adverse event (MACE) is defined as a composite of cardiac death, TVMI, target vessel thrombosis (TVT) or ischemia-driven target vessel/lesion revascularization (TLR/TVR).

The primary outcomes were the peri-procedural MACE, one-year cumulative MACE and angiographic LLL. The secondary outcomes were each component of MACE, MLD and BRS.

Data were expressed as the mean $\pm$ SD for continuous variables or as frequency (%) for discrete or categorical variables. To compare differences between groups, Student’s t test was used for continuous variables, and the chi-square or Fisher’s exact test was used for the discrete variables. A p value of 0.05 was considered statistically significant.

All analyses were performed with SSPS (version 20.0, IBM Corp., Chicago, IL, USA).

The clinical characteristics were balanced between the two groups (Table 1). The use of aspirin, clopidogrel or ticargrelor was similar in the groups regardless of the more frequent use of ticargrelor or less frequent use of clopidogrel in the DCB-based group.

No difference was observed in lesion features between the two groups, especially in bifurcation angulation, branch stenotic severity and lesion length between the two groups, and the proportion of true BLs (100% vs. 100%) and complex BLs (20.0% vs. 23.3%) were similar between the DCB- and the DST-based groups (Table 1).

Procedural data are shown in Table 2. As the DCB-based strategy is a preset combined approach, there was more frequent use of scoring or cutting balloons (MV: 50% vs. 6.7%, p = 0.000; SB: 63.3% vs. 6.7%, p = 0.000) and GP IIb/IIIa inhibitors (60% vs. 6.7%, p 0.001) were observed in the DCB-based group. The length of stenting or DCB angioplasty for both branches was comparable between the groups with less final kissing dilation (26.7% vs. 100%, p = 0.000) in the DCB-based group. Although dissection Type C occurred more frequently in the DCB-based group (MV: 26.7% vs. 3.3%, p = 0.030; SB: 36.7% vs. 6.7%, p = 0.012), there was neither flow-limiting dissection and the associated events nor requirement of bailout stenting during the procedures were noted in the DCB-based group. Optical coherence tomography (OCT) showed that these dissections were minor with an arc 60°, a length of 2 mm and limited to the intima (Fig. 3). The rate of immediate angiographic success defined by residual stenosis 20% was lower in both branches (MV: 46.7% vs. 96.7%, p 0.001; SB: 33.3% vs. 80.0%, p 0.001) in the DCB-based group, but the rate of immediate residual stenosis $>$30% was low and similar in both branches (MV: 6.7% vs. 3.3%, p = 1.000; SB: 10.0% vs. 3.3%, p = 0.605) between the groups.

Fig. 3.

Healing of non-flow-limiting dissection during follow-up. CAG and OCT show a typical true BL affected LAD-D pre-procedurally (A,B,C) with lipid-rich plaque in LAD (B) and D (C), several minor dissections observed post-procedurally (D,E,F) in LAD (E) and D (F) and no more dissections found at 1-year follow-up (G,H,I) in the corresponding site. CAG, coronary angiography; D, diagonal artery; LAD, left anterior descending artery; OCT, optical coherence tomography; BL, bifurcation lesion.

QCA data are listed in Table 3. Compared to the DST-based group, the DCB-based group had a reduced LLL in both branches (MB: 0.05 $\pm$ 0.24 mm vs. 0.25 $\pm$ 0.35 mm, p = 0.013; SB: –0.02 $\pm$ 0.19 mm vs. 0.11 $\pm$ 0.15 mm, p = 0.005). In line with LLL, there were similar MLD and DSP and BRS were observed in both branches between the groups.

No patients were lost to follow-up. The rates of peri-procedural MACEs (0.0% vs. 0.0%, p = 1.000) and one-year cumulative MACEs driven all by TLR/TVR (6.70% vs. 13.30%, p = 0.667) were similar without death and TVT between the DCB- and the DST-based groups (Table 4). The occurrence of post-procedural troponin elevation of $\ge$5 $\times$ URL was similar between the groups (16.7% vs. 20.0%, p = 1.000).