This single-center, retrospective study was performed in the Shanghai Chest Hospital, Shanghai Jiao Tong University School of Medicine, according to the principles of the Declaration of Helsinki and local regulations. This study was approved by the Ethics Committee of Shanghai Chest Hospital, Shanghai Jiao Tong University School of Medicine (ethics ID: IS2128). Written informed consents were waived.

We retrospectively screened adult patients (>18 years old) who (1) were admitted to the Department of Cardiology, Shanghai Chest Hospital, Shanghai Jiao Tong University School of Medicine from January 2012 to May 2019, (2) received the one-stent strategy for de novo non-LM coronary bifurcation lesions, and (3) had documented coronary angiographic images at the index procedure and at 6- to 24-month angiographic follow-up. The de novo non-LM coronary bifurcation lesion was defined as a de novo coronary artery narrowing occurring adjacent to or involving significant SB origin (16), whose site was not at the distal left main coronary artery. Significant SB selection was based on the European Bifurcation Club consensus definition (a branch that the operator would not want to lose in the global context of an individual patient, including symptoms, ischemia location, branch responsible for symptoms or ischemia, viability, and collateralizing vessel) (16), but not the reference SB diameter in order to avoid missing visually small SBs with clinically significance. The mean reference SB diameter was 1.6 mm in this study. Exclusion criteria were ST-elevation myocardial infarction (STEMI), non-ST-elevation myocardial infarction (NSTEMI), cardiomyopathy, myocarditis, valvular heart disease, chronic total occlusion, left main artery lesion, thrombus-containing lesion, and prior coronary artery bypass graft. Angiographic inclusion criterion was post-procedural SB TIMI flow grade 3. Angiographic exclusion criteria included (1) lesions without proper angiographic images for quantitative coronary angiography (QCA) or QFR analysis, (2) MV TIMI flow grade <3 at 6- to 24-month angiographic follow-up, and (3) SB lesion length that cannot be clearly measured.

The PCI strategy and instrumentation use in all the cases were at the discretion of the interventional operators, including coronary stent selection. Coronary angiography was conducted in the conventional manner following the current guidelines. Peri-procedural and long-term anti-platelet and anti-coagulant administration was determined based on the operator's discretion and current guidelines. Administration of aspirin 300 mg and clopidogrel 300–600 mg or ticagrelor 180 mg as the loading doses followed by aspirin 100 mg once daily and clopidogrel 75 mg once daily or ticagrelor 90 mg twice daily as the maintenance doses before the procedure was mandatory. Life-long aspirin (100 mg once day) was prescribed to all patients except for those who had aspirin intolerance. Clopidogrel 75 mg once daily or ticagrelor 90 mg twice daily administration for at least 12 months after procedure was recommended to all patients.

Clinical data were acquired through the medical chart review. A pre-specified data collection form was used to record age, sex, body mass index, diagnosis (stable angina/silent ischemia or unstable angina), co-morbidities (hypertension, diabetes mellitus, hyperlipidemia), smoking status, medical history [prior myocardial infarction (MI), prior PCI), low-density lipoprotein cholesterol, and left ventricular ejection fraction assessed by transthoracic echocardiogram.

The documented cineangiograms were used to review the baseline lesion and procedural characteristics. For the lesion characteristics, bifurcation location, bifurcation type classified using Medina criteria, true bifurcation (defined as Medina 1,1,1, 1,0,1, or 0,1,1) (17), plaque location, coronary calcification, angulation, and irregular plaque were recorded. Moderate calcification was defined as radiopacities noted only with the cardiac motion before contrast injection. Severe calcification was defined as radiopacities noted without cardiac motion before contrast injection generally compromising both sides of the arterial lumen (18). Moderate–severe angulation was defined as lesion angulation >45° (19). Irregular plaque was defined as the plaque with visually unsmooth surface (20). For the procedural characteristics, MV characteristics (i.e., dissection before MV stenting, stent type, stent diameter, stent length, maximal balloon pressure, and maximal balloon diameter) and SB [i.e., SB pre-dilation, kissing balloon inflation (KBI) before MV stenting, dissection before MV stenting, SB TIMI flow grade before MV stenting, SB protection use including jailed wire technique, and jailed balloon technique, SB TIMI flow grade after MV stenting, SB opening after MV stenting, drug-coated balloon use, final KBI, and final SB TIMI flow grade] were recorded.

Clinical outcomes data were also collected by reviewing the medical charts and the patients or their families were contacted for verification via telephone. The Academic Research Consortium-2 consensus was accordingly used to define death, MI, target vessel revascularization, and stent thrombosis (21).

Post-procedural SB μQFR was analyzed. The AngioPlus Core software (Pulse Medical, Shanghai, China) were used to perform offline QFR analyses following standard operation procedures as described in previous studies (13, 22). QFR analyses were conducted by two independent, well-trained investigators (YJ and WT) who were blinded to patient clinical characteristics and follow-up.

QCA analysis was performed offline following the standard analysis procedure using a computer-based system dedicated to bifurcation (QAngio XA, version 7.3, Medis, Leiden, Netherlands) (23). The bifurcation lesions were divided into the proximal MV, distal MV, and SB segments. Bifurcation angle, lesion length, minimal lumen diameter (MLD), reference vessel diameter (RVD), diameter stenosis (DS), acute gain, and late loss were detected. The bifurcation angle was defined as the angle between the distal MV and SB segments. Binary restenosis was defined as QCA DS >50% (24). QCA analyses were conducted by two independent well-trained investigators (DW and YZ) who were blinded to patient clinical characteristics and follow-up.

Coronary blood flow was quantified with the TIMI grading system according to a previous study (25) and determined by visual estimation. TIMI flow grade were assessed by two independent well-trained investigators (KX and WZ) who were blinded to patient clinical characteristics, QCA analyses, and follow-up. The κ coefficient was used to detect the inter-operator variability.

Data for clinical characteristics were presented on a per-patient basis and on a per-vessel basis for the remaining calculations. All the included lesions were divided into three tertile groups according to the post-procedural SB μQFR and then included in analyses. Continuous data were present as the mean ± standard deviation and compared using the one-way analysis of variance (ANOVA) tests. Categorical data were present as the number (%) and compared using the chi-square tests. Receiver operating characteristic (ROC) curve analyses were used to determine the ability of post-procedural SB μQFR and SB DS to distinguish between SBs with and without SB TIMI flow grade ≤1 at angiographic follow-up, respectively, and to identify the optimal cut-off value of post-procedural SB μQFR that provided the greatest sum of specificity and sensitivity. Logistic regression models were adopted to detect the association of post-procedural SB μQFR with the incidences of long-term SB TIMI flow grade ≤1. Characteristics of clinical information (age, sex, hypertension, hyperlipidemia, diabetes, smoking status), lesion (true coronary bifurcation, bifurcation location), and procedure (stent diameter, stent type, MV RVD, and SB RVD) were included in multivariable regression analyses. DeLong's method was used to compare ROC curves between post-procedural SB μQFR and SB DS. Subgroups analyses were performed according to baseline SB RVD: ≤1.5, 1.5–2.0 mm, and >2.0 mm. Results were presented as the odds ratios (ORs) with 95% confidence intervals (CIs).

All analyses were conducted with Stata/SE software version 17.0 (Stata Corp, College Station, TX, USA) and GraphPad Prism version 9.4.0 (GraphPad Software, San Diego, CA, USA). A two-sided P-value of <0.05 was considered statistically significant.

We screened 554 non-LM coronary bifurcation lesions receiving one-stent strategy in 280 patients, and finally included 313 lesions in 146 patients who met all the inclusion criteria and had none of any exclusion criteria in this study (see Figure 1).

Clinical characteristics are detailed in Table 1. The mean age was 64.5 years. Of them, 111 (76.0%) patients were male. Lesion characteristics are detailed in Table 2. A total of 227 (72.5%) were located at the left anterior descending artery (LAD) and 53 (16.9%) were true bifurcation lesions. Procedural characteristics are detailed in Table 3. The stent diameter was 2.98 mm ± 0.36 mm, and stent length was 29.33 mm ± 6.91 mm.

Baseline clinical characteristics of patients with and without long-term SB TIMI flow grade ≤1 are shown in Supplementary Table S1. All the characteristics were comparable. Lesion characteristics in lesions with and without long-term SB TIMI flow grade ≤1 are shown in Supplementary Table S2. More complex lesions and more true bifurcation lesions were observed in lesions with long-term SB TIMI flow grade ≤1. Procedural characteristics in lesions with and without long-term SB TIMI flow grade ≤1 are detailed in Supplementary Table S3. More lesions with impaired SB TIMI flow before and after MV stenting were observed in lesions with long-term SB TIMI flow grade ≤1.

The distribution of pre- and post-procedural SB μQFR is presented in Figure 2. In the entire cohort, the median and interquartile post-procedural SB μQFR range were 0.89 and 0.15 (0.81–0.96). All the included lesions were divided into three tertile groups according to the post-procedural SB μQFR: low tertile group (SB μQFR ≤ 0.84, n = 108), middle tertile group (0.84 < SB μQFR ≤ 0.93, n = 106), and high tertile group (SB μQFR > 0.93, n = 99).

Pre-stenting, post-stenting, and follow-up μQFR in MV and SB in the three groups are shown in Tables 4, 5, respectively. The post-stenting and follow-up SB μQFR both significantly increased across the tertiles of post-procedural SB μQFR (all P < 0.05). The post-stenting and follow-up SB μQFR were both found to significantly increase across the tertiles of post-procedural SB μQFR (all P < 0.05), but the absolute difference among groups was small.

Pre-PCI, post-PCI, and follow-up μQFR in lesions with and without long-term SB TIMI flow grade ≤1 are shown in Supplementary Table S4. Pre-stenting, post-stenting, and follow-up μQFR were significantly lower in lesions with long-term SB TIMI flow grade ≤1 than those with long-term SB TIMI flow grade >1 (all P < 0.05). No significant change of μQFR between post-stenting and follow-up was observed.

The results of QCA analyses in MV and SB are detailed in Tables 4, 5, respectively. The lesion length, DS, and incidences of SB binary restenosis at angiographic follow-up decreased significantly across the post-procedural SB μQFR tertiles (all P < 0.05). SB RVDs among the three groups were comparable.

QCA analyses in lesions with and without long-term SB TIMI flow grade ≤1 are detailed in Supplementary Table S5. Pre-PCI and follow-up distal % DS; post-PCI and follow-up SB MLD; and pre-PCI, post-PCI, and follow-up SB % DS were significantly different between lesions with and without long-term SB TIMI flow grade ≤1 (all P < 0.05). Binary restenosis was more observed in lesions with long-term SB TIMI flow grade ≤1 than those without long-term SB TIMI flow grade ≤1 (62.5% vs. 10.4%, P < 0.0001).

A total of 8 (2.6%), 16 (5.1%), 61 (19.5%), and 228 (72.8%) SBs had a TIMI flow grade of 0, 1, 2, and 3 at follow-up, respectively. Supplementary Table S6 presented the SB TIMI flow grade in the three groups at angiographic follow-up. The incidence of SB TIMI flow grade ≤1 and ≤2 both numerically decrease across the post-procedural SB μQFR tertiles (SB TIMI flow ≤ 1: P = 0.086; SB TIMI flow ≤ 2: P = 0.052; Figure 3).

Also, collateral flow from other coronary artery was observed in six (75.0%) SBs with long-term SB TIMI flow of 0, six (37.5%) with long-term SB TIMI flow of 1, two (3.3%) with long-term SB TIMI flow of 2, and zero with long-term SB TIMI flow of 3.

The κ coefficient for inter-operator variability in assessing SB TIMI flow grade was 0.78, and the κ coefficient for inter-operator variability in assessing SB TIMI flow grade ≤1 or >1 was 0.95.

Figure 4 shows the ROC curve. The area under the curve was 0.6673 (95% CI, 0.5610–0.7737; P = 0.0064), and the optimal cut-off value of post-procedural SB μQFR for identifying long-term SB TIMI flow grade ≤1 was 0.77, with the specificity of 37.50% (21.16%–57.29%) and the sensitivity of 87.20% (82.85%–90.57%). In the univariable logistic regression model, post-procedural SB μQFR ≤0.77 was associated with a 3.25-fold increased risk in long-term SB TIMI flow grade ≤1. After adjusting the characteristics of patient (age, gender, hypertension, hyperlipidemia, diabetes, smoking status), lesion [true coronary bifurcation, LAD bifurcation], and procedure (stent diameter, stent type), post-procedural SB μQFR ≤0.77 was independently associated with a 2.57-fold increased risk (adjusted OR, 2.57; 95% CI, 1.02–7.25; P = 0.045) in long-term SB TIMI flow grade ≤1 in a multivariable logistic regression model (Table 6).

Supplementary Table S7 detailed the ROC analyses results of post-procedural SB μQFR and SB DS to identify the SB TIMI flow grade ≤1 at angiographic follow-up in the all included lesions and in groups with different SB RVD. No significant differences in diagnostic accuracy between post-procedural SB μQFR and SB DS were observed in all included lesions and in groups with different SB RVDs.

The follow-up duration was 12.3 ± 3.0 months. No death or stent thrombosis occurred in the overall population. MI occurred in two (1.5%) patients. A total of 14 (9.6%) patients received target vessel revascularization.

As a pressure wire–free technique, μQFR could overcome the technical difficulty of crossing a pressure wire to the stent struts, avoid FFR-related peri-procedural complications, and reduce the upfront cost of pressure wire in all the patients when assessing SB compromise. Also, μQFR can be performed in patients allergic to vasodilators or ineligible due to the lesion characteristics of those who could not receive FFR assessment. Thus, μQFR can be used to optimize PCI strategy in more patients than FFR. Moreover, μQFR could improve clinical feasibility in SB evaluation as compared with the original QFR version because it requires two satisfied angiographic views acquired with a minimal of 25° in separation. The μQFR is a safe and easy approach to assess SB functional compromise in routine practice and is thus potentially useful for optimizing non-LM coronary bifurcation lesion interventional strategy in routine clinical practice. This requires further investigation. Given the complexity of coronary bifurcation intervention, the decision-making on further SB intervention after MV stenting may need to involve additional information under consideration, such as clinical importance, technical difficulty, and complication risk.

First, as a retrospective study, the incidences of different SB TIMI flow grades at follow-up in non-LM coronary bifurcation lesions receiving the one-stent strategy might be biased by patient selection. Large-scale prospective studies with controlled angiographic follow-up may provide more useful information. Second, the μQFR computation was performed offline. Online μQFR performance requires further investigation. Third, historical angiographic image records for analyses were used, which might challenge the μQFR computation due to the suboptimal projections for bifurcation lesions and images quality. A previous study revealed that μQFR had high diagnostic accuracy in estimating FFR in suboptimal projections (13). Also, agreement between QFR and FFR was observed to be good in vessels with angiograms acquired under low x-ray frame and pulse rate setting (7.5 fps and 7 pps modes) (39). Thus, μQFR computation can be used in poor-quality angiograms. Fourth, the mean SB RVD was 1.6 mm, smaller than most of previous studies on coronary bifurcation lesions. Thus, the incidence of SBs with TIMI flow grade ≤2 might be higher than that in previous studies (40). Fifth, the baseline proximal MV RVD increased across the tertiles of post-procedural SB μQFR in this study, from 2.87 mm in the low tertile group to 3.10 mm in the high tertile group. One explanation is that lesions in the low tertile group had more diffused coronary atherosclerotic plaques affecting accurate RVD assessment. The absolute difference in baseline RVD of proximal MV between the low and high tertile groups was low (0.23 mm), which might have little clinical significance. We had also included the RVDs of both MV and SB into multivariable models for adjustment. Results showed that no association existed between MV RVD and long-term SB TIMI coronary blood flow, and the association between post-procedural SB μQFR and long-term SB TIMI coronary blood flow was not influenced by the MV RVD. Sixth, the specificity of μQFR <0.77 as 37.50% was low. Considering that the cut-off value of SB μQFR is used for identifying SB at higher risk in long-term impaired blood flow and might be helpful to determine the SB reintervention immediately after index procedure, the sensitivity might be more clinically important. In this study, the sensitivity of μQFR <0.77% as 87.20% was satisfactory. Also, other cofounders may be also important for long-term SB TIMI flow grade. Seventh, since this study was a retrospective study, we did not record data regarding angina severity. Future prospective studies are required to address these questions.