The Clinical Outcome of FFR-guided Revascularization Strategy of Coronary Lesions (HALE-BOPP) study is an investigator-initiated, multicentre, international prospective study conducted in ten hospitals between Italy and the United Kingdom. The study organization and the participating centres are listed in the supplemental online. The study consecutively enrolled all patients who underwent FFR measurement on at least one vessel with COMET® wire (H74939359310, Boston Scientific, Natick, MA, USA) (Fig. 1). Exclusion criteria were life expectancy of less than one year because of known non-cardiovascular comorbidity, inability to guarantee clinical follow-up, and unwillingness to provide written informed consent. Patients with prior coronary artery bypass (CABG) and chronic total occlusion (CTO) were also excluded.

Fig. 1.

Study flow-chart. FFR, fractional flow reserve; CABG, coronary artery graft bypass; CTO, chronic total occlusion; TVR, target vessel revascularization.

All vessels showing a lesion with a diameter stenosis (DS) $\ge$50% (by visual estimation) were of interest to the study. The final decision to measure FFR or to proceed with an angio-based revascularization or deferral was left to the operator. The protocol strongly suggested that vessels with the culprit lesion of ACS or with lesions showing DS $\ge$90% should be treated with percutaneous coronary intervention (PCI) and second-generation drug-eluting stent (DES) avoiding FFR measurement. For all the other vessels showing at least one lesion greater than 50%, the protocol strongly suggested performing an FFR evaluation to guide revascularization. An FFR assessment was performed according to the procedures previously described [9]. FFR requires the use of maximal hyperaemia, which can be induced by both systemic and intracoronary administration of adenosine. The FFR was calculated by evaluating the ratio between the proximal aortic pressure and distal coronary pressure during the steady state of maximal hyperaemia [10]. Potential pitfalls related to wire (i.e., drift, coronary dissection, coronary perforation, inability to cross the stenosis, etc.) were prospectively recorded. After each measurement, pressure-wire pullback to check for pressure-drift was strongly recommended. A drift-value from 0.96–1.04 was accepted. For FFR values of 0.76–0.84, a drift with a narrower range of 0.98–1.02 was accepted [11]. The FFR value was considered flow-limiting (positive) if $\le$0.80 and coronary revascularization was mandated by protocol. Conversely, an FFR value $>$0.80 was considered not flow-limiting (negative), and coronary revascularization had to be deferred. In accordance with decision-making workflow and treatment, vessels were classified into three subgroups: (i) angio-revascularized, (ii) FFR-revascularized, (iii) FFR-deferred. A PCI procedure was performed using standard materials and techniques. Subsequently, patients received standard medical therapy according to the current guidelines [12], with particular emphasis on achieving the recommended targets of low-density lipoprotein cholesterol (LDL cholesterol). Clinical follow-up occurred at 1, 6, and 12 months and annually thereafter. For the present analysis, follow-up was censored in March 2022 or at the time of death.

All baseline, clinical, lesion, and outcome data were prospectively collected using a dedicated electronic case report form (eCRF). The specialized personnel at each centre followed this paradigm. Members of the academic coordinating centre (University of Ferrara, Ferrara, Italy) periodically performed monitoring and verification of data in the Italian hospitals. Members of the contract research organization GBPharma (Pavia, Italy) monitored and verified the data of the United Kingdom centre. Angiograms and FFR traces were prospectively analysed at an independent core laboratory (University of Ferrara, Ferrara, Italy) without knowledge of the patient’s outcomes. Angiographic analyses were carried out for all lesions with a 50% or greater DS in each epicardial vessel and side branch that were 1.5 mm or larger in diameter, using an automated edge-detection algorithm (QAngio XA 7.3, Medis Medical Imaging Systems, Leiden, Netherlands). FFR traces were reviewed for the quality in hyperaemia induction, drift check and consistency for the FFR value reported in the eCRF.

The main analysis was carried out at vessel-level. The study endpoint was the target vessel failure (TVF), defined as the cumulative occurrence of cardiac death, target vessel myocardial infarction and ischemia-driven target vessel revascularization. The prespecified time-point of the primary outcome was at 1 year. Due to the limited number of adverse events and to better describe the natural history of coronary lesions whose treatment was deferred based on the FFR results, the present analysis reports data outcome from the longest-term follow-up. Adverse events are defined in the supplemental online and were adjudicated by a clinical events committee that reviewed original source documents. In case of repeated adverse events, the first one that occurred was the one considered. In addition, the committee assigned each event to a specific coronary vessel based on the available information (i.e., electrocardiogram, cardiac biomarkers, echocardiography, coronary artery angiography). In the case of cardiovascular death in patients with multiple study vessels, the event was assigned to each vessel [13].

To support and confirm the findings of the vessel-level analysis and allow comparison with previous studies, data are also analysed at patient-level. According to the vessel status, patients were defined: (i) negative-FFR if all vessels interrogated with FFR were classified as FFR-deferred, (ii) positive-FFR if at least one interrogated vessel was classified as FFR-revascularized [6]. The composite study endpoint included cardiac death, myocardial infarction, and ischemia-driven coronary revascularization.

Starting from previous similar studies [14, 15, 16], we expected a 1-year incidence of the endpoint at around 5% in the FFR-deferred vessels. Setting a tolerance margin at around 1.5%, at least 811 patients with at least one FFR-deferred vessel were needed. Continuous data were tested for normal distribution with the Kolmogorov-Smirnov test. Normally distributed values were presented as mean $\pm$ SD, otherwise, the median value and interquartile range (IQR) were used. Categorical variables were summarised in terms of counts and percentages. For the comparison between groups, t-test, Mann-Whitney U test and Pearson’s $\chi$${}^2$ test were applied as appropriate. Unadjusted survival was examined with Kaplan-Meier survival curves and the log-rank test. Overall, we performed evaluations “per vessel”, but these evaluations were clustered by patient. In this way, the assumption of independence between vessels was violated. Therefore, to adjust for this clustering effect, we used multilevel modelling and shared frailty Cox proportional hazards regression as the primary model [17, 18]. Cox regression analysis with interaction testing was performed to determine whether the effect of the FFR-based deferral strategy (vs. coronary revascularization) on the primary endpoint was consistent across different subgroups of clinical interest. To support the findings of the vessel-level analysis, the analyses were also repeated at patient-level. All analyses were performed with Stata version 13.1 (StataCorp LP, College Station, TX, USA).

Operators proceeded with angio-guided revascularization in 760 vessels (angio-revascularized, Table 2, Supplementary Table 2). The reasons behind the use of the angio-guided approach as reported by the operators are recorded in Supplementary Table 3. Conversely, in 1662 vessels, the FFR was assessed to guide revascularization (FFR-guided vessels, Table 2). Overall, in 1126 (67%) vessels, the treatment was deferred (FFR-deferred), whereas 536 (33%) vessels were treated with PCI (FFR-revascularized) (Table 2). During the FFR assessment, 14 (0.8%) cases of wire-related pitfalls were observed (drift n = 12, inability to cross the lesion n = 2). No perforations or dissections were reported. Adenosine was administered intravenously in 1459 (87.9%) measurements. Fifty-six (3.3%) patients complained adenosine-related side effects (significant dyspnoea n = 51, hypotension n = 3, marked bradycardia n = 2). The median FFR value in the FFR-revascularized group was 0.74 [0.70–0.78], while it was 0.88 [0.84–0.92] in the FFR-deferred group.

In order to better characterise the vessels under consideration, the characteristics and annual event rate of the 121 vessels that were deferred on the basis of the angiographic evaluation (Angio-deferred) are also reported in the online Supplementary Table 1.

At 1-year, 13 (1.1%) cardiovascular deaths, 9 (0.6%) target vessel myocardial infarction (MI), and 15 (1.3%) ischemia-driven target vessel revascularizations occurred in the FFR-deferred vessels. TVF occurred in 29 (2.5%, 95% CI: 1.9%–3.1%) vessels, which was significantly inferior to the prespecified primary endpoint estimation (from 3.5% to 6.5%). During a median follow-up of 3.6 [2.5–4.7] years, 31 (2.8%) cardiovascular deaths, 19 (1.7%) target vessel MI and 38 (3.4%) ischemia-driven target vessel revascularizations occurred. Altogether, TVF occurred in 68 (6%) vessels. The overall incidence rate of TVF was 0.024 (95% CI: 0.019–0.031) lesion/year. The univariate associations between potential predictor covariates and TVF are shown in Supplementary Table 4. The final predictors of TVF in FFR-deferred vessels were CKD (hazard ratio [HR]: 2.74, 95% CI: 1.22–6.12), multivessel disease (HR: 3.69, 95% CI: 1.22–6.12) and LVEF (0.95, 95% CI: 0.92–0.98).

At the longest-term follow-up, the TVF incidence rate lesion/year and the unadjusted TVF cumulative occurrence in the FFR-revascularized vessels were 0.029 (95% CI: 0.02–0.03) and 7%, respectively. They were significantly lower than those of the angio-revascularized vessels (0.049, 95% CI: 0.040–0.061, p = 0.0001 and 11.7%, HR: 0.58, 95% CI: 0.44–0.76, p = 0.0001, respectively).

TVF occurred in 15 (12.4%) angio-deferred vessels. The incidence rate of TVF was 0.055 (95% CI: 0.033–0.092). As compared to angio-deferral, FFR-deferral was an independent protective factor for TVF also after correction for potential patient’s and vessel’s confounding factors (HR: 0.38, 95% CI: 0.18–0.79, p = 0.01). Similarly, compared to revascularized vessels (angio- and FFR-revascularized, target vessel revascularization [TVR] incidence rate lesion/year 0.040, 95% CI: 0.033–0.047, unadjusted TVF cumulative occurrence 9.4%), FFR-deferral was independently associated with a better outcome also after correction for potential patient’s and vessel’s confounding factors (HR: 0.69, 95% CI: 0.48–0.98, p = 0.044). This observation was consistent across subgroups of clinical interest (Fig. 2). Compared to angio-guidance (angio-deferred + angio-revascularized vessels), FFR-guidance (FFR-deferred + FFR-revascularized vessels) was associated with a lower occurrence of TVR (Fig. 3) and with a significant reduction of risk of TVF (group HR: 0.58, 95% CI: 0.44–0.76, p = 0.0001). The predictors of TVF in all study vessels were female sex (0.41, 95% CI: 0.20–0.84, p = 0.014), CKD (2.15, 95% CI: 1.16–3.96, p = 0.014), ACS (2.75, 95% CI: 1.55–4.86, p = 0.0001), de novo lesion (0.37, 95% CI: 0.17–0.80, p = 0.012) (Supplementary Table 5).

Fig. 2.

Subgroup analysis in the comparison of FFR-deferral vs. revascularization. HR, hazard risk; CKD, chronic kidney disease; CCS, chronic coronary syndrome; ACS, acute coronary syndrome; LVEF, left ventricular ejection fraction; LM, left main; LAD, left anterior descending artery; RCA, right coronary artery; LCx, left circumflex artery; FFR, fractional flow reserve.

Fig. 3.

Cumulative survival free from target vessel failure in FFR-guided vessels vs. Angio-guided vessels. FFR, fractional flow reserve; HR, hazard risk.

Overall, 814 (62%) and 491 (38%) patients were classified as negative-FFR and positive-FFR, respectively (Fig. 1). At the longest available follow-up, 40 (3.0%) patients died of a cardiovascular cause, 63 (4.8%) experienced MI and 110 (8.4%) underwent ischemia-driven coronary revascularization. The TVF incidence rate lesion/year was significantly lower in the negative-FFR patients compared to the positive-FFR ones (0.027, 95% CI: 0.020–0.035 vs. 0.044, 95% CI: 0.033–0.058, p = 0.021). Also, the unadjusted TVF cumulative occurrence was lower in negative-FFR patients (HR: 0.63, 95% CI: 0.42–0.93, p = 0.021) (Supplementary Fig. 1). After correction for potential confounding factors, female sex (HR: 0.51, 95% CI: 0.28–0.91), diabetes mellitus (HR: 0.49, 95% CI: 0.26–0.91), LVEF (HR: 0.96, 95% CI: 0.94–0.99) and multivessel disease (HR: 4.71, 95% CI: 2.37–9.59) were independent predictors of TVF (Supplementary Table 6). We found a significant interaction between ACS and CCS patients (0.79, 95% CI: 0.53–1.16 vs. 0.75, 95% CI: 0.40–1.39, p for interaction = 0.007) and between patients with or without diabetes (0.32, 95% CI: 0.15–0.68 vs. 0.99, 95% CI: 0.68–1.44, p for interaction = 0.017) (Supplementary Fig. 2).