The study protocol was approved by the ethics committees of the participating centers (IRB no: CNUH-2018-143) and adhered to the principles of the Declaration of Helsinki. All patients provided written informed consent before inclusion in the registry.

The FRAME-AMI trial design and its principal results have been discussed in detail previously, specifically in the appendix of the primary publication (14).

Briefly, the FRAME-AMI trial was an open-label, multicenter, randomized study that contrasted the outcomes of FFR-guided vs. angiography-guided PCI for non-culprit arteries in AMI with MVD. The culprit artery was defined as the artery related to AMI, and it was determined by the operator based on angiography and/or other modalities, including electrocardiogram, echocardiography, or intravascular imaging, if indicated. Other coronary arteries were considered non-culprit arteries. MVD was defined as stenosis > 50% in a non-culprit vessel with a diameter ≥ 2.0 mm, as visually estimated. The study included patients aged over 18 years diagnosed with AMI. Primary PCI for the culprit artery was undertaken within 12 h of symptom onset in cases of ST elevation myocardial infarction (STEMI), and within 72 h in non-ST elevation myocardial infarction (NSTEMI) cases. The trial protocol was approved by the institutional review board at each participating site.

Exclusion criteria included cardiogenic shock at presentation, unprotected left main coronary artery disease (stenosis > 50% by visual estimation), non-culprit arteries not amenable for PCI, severe stenosis with thrombolysis in myocardial infarction (TIMI) flow II or less, or chronic total occlusion. Further details of inclusion and exclusion criteria have been previously described (14).

Following the culprit artery PCI, evaluation and PCI for non-culprit arteries were conducted either immediately or at a later stage, based on the operator's discretion. While the interval between index and staged PCI was not pre-determined, staged PCI for non-culprit arteries was carried out during the index hospitalization. CR was defined when all non-culprit arteries were treated with PCI according to predefined criteria based on FFR or angiographic guidance as described below.

In the FFR group, FFR was carried out in all non-culprit arteries with a lesion(s) presenting > 50% stenosis on visual estimation, and PCI for non-culprit arteries was performed only when the FFR result was 0.80 or lower. In the angiography group, lesions with a diameter stenosis > 50% were treated with PCI. All medical treatments adhered to the current guidelines (15), and dual antiplatelet treatment was maintained for a minimum of 12 months.

This study's objective was to assess the impact of the timing of PCI for non-culprit arteries in AMI with MVD. To this end, we conducted a post-hoc analysis of the FRAME-AMI trial. Participants were categorized into immediate and staged CR groups based on the timing of evaluation and PCI for non-culprit arteries. All staged PCIs were performed during the index hospitalization. The flow of analyses is depicted in Figure 1. After excluding patients treated for non-culprit arteries in violation of the treatment protocol, comparisons were made for all AMI patients and separately for STEMI and NSTEMI cohorts.

The primary endpoint of this study was a composite of all-cause death, myocardial infarction (MI), or any repeat revascularization. Secondary endpoints encompassed each component of the primary endpoint, cerebrovascular accident (CVA), definitive or probable stent thrombosis, and contrast-induced nephropathy (CIN). The definition of death adhered to the guidelines set by the Academic Research Consortium (16). All deaths were presumed to be cardiac unless a distinct non-cardiac cause could be established. MI was defined in accordance with the third universal definition (17), which includes recurrent or procedural-related MI and thus encompasses periprocedural MI. The definition and classification of stent thrombosis adhered to the Academic Research Consortium's guidelines (16). More detailed definitions of clinical outcomes have been previously described in the appendix of the main paper (14).

Categorical variables were analyzed using the chi-square or Fisher's exact tests. Continuous variables are reported as the mean ± SD or median with interquartile range and were analyzed using the Student's t-test or the Wilcoxon rank-sum test. A two-tailed p-value less than 0.05 was deemed to indicate statistical significance.

Kaplan–Meier analyses were conducted to compare the primary and secondary endpoints in groups with varying treatment strategies. The log-rank test was used to assess differences in survival between the groups.

Cox-proportional hazard regression models were employed to ascertain whether staged or immediate CR constitutes an independent predictor of clinical outcomes. Variables deemed significant in the univariable analysis (p < 0.1) or those with clinical significance for each outcome were included in the multivariable analysis. Multivariable Cox-proportional hazard models were prepared with propensity score (PS)-based overlap weighting as described below.

PSs were calculated to mitigate the confounding effects of variable differences in the variable distributions between the two groups and overlap weighting (OW) (18, 19) was applied. PSs were calculated for the revascularization strategies in the analyses (Supplementary Figures S1–S3) using the PSweight package. Several variables were adjusted for the timing of CR including age, sex, hypertension, systolic blood pressure (SBP), diastolic blood pressure, diabetes mellitus (DM), current smoking, door to balloon time, three-vessel disease, left main disease, location of the culprit lesion, transradial approach, use of a glycoprotein IIb-IIIA inhibitor, direct stenting for the culprit lesion, image-guided PCI for culprit vessels, creatinine, total numbers of culprit lesion(s) and non-culprit lesions, left ventricular ejection fraction, FFR guidance for non-culprit lesion, and STEMI in cases of all AMI. Histograms for PS and plots for PS density with weighting are provided for AMI, STEMI, and NSTEMI (Supplementary Figures S4–S6, respectively). Furthermore, subgroup analysis for the primary endpoint was performed for major clinically significant variables.

All statistical analyses were conducted using the R statistical package (version 4.2.0; R Foundation for Statistical Computing, Vienna, Austria).

The baseline clinical characteristics of the patients are presented in Table 1.

Several statistically significant differences were observed between the groups. The immediate CR group had a lower proportion of males, with higher SBP, more frequent DM with higher hemoglobin A1C, lower creatinine, and higher left ventricular ejection fraction.

The procedural profiles are compared in Table 2. Immediate CR was less frequently performed in patients with STEMI (33.1% vs. 66.8%, p < 0.001). The distribution of culprit vessels also differed (p < 0.001), with a higher proportion of left anterior descending (LAD) (40.4% vs. 26.4%) and a lower proportion of right coronary artery (RCA) (33.4% vs. 54.5%), and less frequent three-vessel disease (31.6% vs. 48.6%, p < 0.001) in the immediate CR group. Pre-PCI diameter of the stenosis and the FFR value for non-culprit arteries were comparable between the two groups, but the total length of the non-culprit lesions was longer in immediate CR.

Procedural differences during PCI between the two groups were noted. The immediate CR group showed a lower rate of transfemoral approach (9.7% vs. 25.5%, p < 0.001) and use of glycoprotein IIb-IIIa inhibitors (15.2% vs. 24.1%, p = 0.012). Image-guided PCI for the culprit artery was more frequent in the immediate CR group (24.9% vs. 16.4%, p = 0.022). However, image guidance for PCI of non-culprit arteries was less frequently used in this group (24.9% vs. 36.8%, p = 0.009). The rate of FFR usage was statistically comparable in both groups.

Although mean FFR value for non-culprit artery was comparable (0.70 ± 0.1), non-culprit arteries with diameter stenosis ranged 70 to 90% was more prevalent in the staged CR group (46.2% vs. 54.3, p = 0.034).

The total number of stents used in the whole PCI procedure was lower in the immediate group (2.3 ± 0.9 vs. 2.5 ± 1.1, p = 0.001), with a shorter total length of stents (34.8 ± 20.0 mm vs. 43.8 ± 21.3 mm, p < 0.001), and a significantly lower number of stents for non-culprit arteries (1.0 ± 0.8 vs. 1.3 ± 0.9, p < 0.001) in the immediate CR group.

The length of hospital stay was not statistically different (3.0 days as median in both groups, p = 0.197). The total rate of in-hospital complications was comparable between groups (3.6% vs. 6.4%, p = 0.206), except for a lower rate of cardiogenic shock in the immediate CR group (0.3% vs. 2.7%, p = 0.036).

The baseline clinical characteristics of the patients are presented in Supplementary Table S1. Patients with STEMI and MVD were less frequently treated with immediate CR than those with NSTEMI (42.5% vs. 75%, p < 0.001). Baseline characteristics in both STEMI and NSTEMI groups were statistically similar except for a higher heart rate and DM prevalence in STEMI (34.9% vs. 21.1%, p = 0.021), and higher SBP in NSTEMI patients treated with immediate CR.

Procedural profiles are compared in Supplementary Table S2. For STEMI patients, the door-to-balloon time (DBT) for a culprit artery was comparable between the two groups (median: 72.5 min vs. 68.0 min, p = 0.113). The distribution of the culprit vessel significantly differed (p = 0.002), with a higher proportion of LAD (50.5% vs. 28.6%) and lower proportion of right coronary artery (RCA) (37.6% vs. 56.5%) in the immediate CR group. The transfemoral approach was less frequently used (11.9% vs. 25.9%, p = 0.009) during immediate CR. In contrast to the similar usage rate of imaging devices for the culprit artery, imaging guidance was less frequently used for non-culprit arteries in the immediate CR group (23.2% vs. 38.1%, p = 0.027). Additionally, a lower number (1.1 ± 0.6 vs. 1.3 ± 0.8, p = 0.023) and shorter total length of stents (32.3 ± 15.0 mm vs. 40.8 ± 20.9 mm, p = 0.001) were used for non-culprit arteries in immediate CR. The incidence of in-hospital complications was comparable.

In NSTEMI patients, the DBT was longer in the immediate CR group (median: 625.5 min vs. 350.0 min, p = 0.001). There was also a difference in the distribution of culprit vessels (p = 0.009) between the two groups; similar to STEMI, LAD was more frequent (35.5% vs. 21.9%) and RCA was less common (31.5% vs. 50.7%) in the immediate CR group. The transfemoral approach was also less frequently used in immediate multivessel PCI (8.6% vs. 24.7%, p = 0.001).

Fewer stents were used (1.0 ± 0.8 vs. 1.4 ± 1.0, p = 0.001) with a shorter total stent length (36.2 ± 22.3 mm vs. 50.1 ± 20.9 mm, p < 0.001) for non-culprit arteries during immediate CR. The incidence of in-hospital complications, including CIN, was comparable.

Clinical outcomes are detailed in Table 3. The results of Kaplan-Meier analysis for the primary endpoints in AMI, STEMI, and NSTEMI are provided in Figure 2.

Among all AMI patients, during the follow-up period (median 3.48 years, 1Q/3Q = 2.68/4.08 years), the primary endpoint occurred in 35 and 22 patients in the immediate and staged CR groups, respectively. All-cause death occurred in 13 and 8 patients in each group. Among them, 11 and 7 cardiac deaths occurred in each group. There was no in-hospital mortality in either group. No other outcomes were significantly different.

In patients with STEMI, composite outcomes occurred in 11 patients in each group and all-cause death occurred in 1 and 4 patients in each group. All mortalities were cardiac-related. In patients with NSTEMI, the primary endpoint occurred in 24 and 11 patients in each group (13.2% vs. 29.2%). All-cause death occurred in 12 and 4 patients in each group (6.0% vs. 20.3%). Other secondary endpoints did not significantly differ. The results of Kaplan-Meier analysis for secondary endpoints in AMI, STEMI, and NSTEMI are provided in Supplementary Figures S7–S9.

The results of the subgroup analysis for the primary endpoint are displayed in Figure 3. Because subgroups according to presentation (STEMI or NSTEMI) were analyzed in detail separately, the subgroups were not included in Figure 3 (interaction p = 0.281). There were no statistically significant differences in HR for the primary endpoint from either immediate or staged PCI for nonculprit arteries in all analyzed subgroups.