Impact of stenosis resistance and coronary flow capacity on fractional flow reserve and instantaneous wave-free ratio discordance: a combined analysis of DEFINE-FLOW and IDEAL

This analysis included patients from two international multi-centre studies on comprehensive invasive physiological stenosis assessment: the IDEAL registry, and the DEFINE-FLOW study (NCT02328820). Rationale and design of DEFINE-FLOW [15] and the results of IDEAL [16] have been published elsewhere. From DEFINE-FLOW, solely measurements approved by the core laboratory were used in this subanalysis, since the required physiological data for hyperaemic stenosis resistance and CFC calculations were only available in cases in which core lab data were reported.

Intracoronary nitroglycerin (100–300 µg) was administered at the beginning of the procedure, and repeated every 30 min if necessary. After diagnostic coronary angiography, a 0.014′′ dual pressure and Doppler flow velocity sensor tipped guidewire (ComboWire XT; Philips Volcano, San Diego, CA) was zeroed to atmospheric pressure, and subsequently calibrated to aortic pressure at the ostium of the guiding catheter. Afterwards, the guidewire was positioned at least five vessel diameters distal to the lesion. After obtaining a stable flow signal, wire position was recorded fluoroscopically and adenosine was administered by an intracoronary bolus injection of 60–150 µg, or intravenous infusion of adenosine at a dose of 140 μg/kg/min to induce hyperaemia [17, 18]. After obtaining the measurements, the guidewire was pulled back to the guiding catheter to assess pressure drift.

FFR was calculated as the mean distal to aortic pressure at peak hyperaemia. CFR was calculated as the ratio of hyperaemic (hAPV) to baseline (bAPV) average peak flow velocity. iFR was calculated by dividing distal resting pressure by aortic pressure in the diastolic wave-free period. Baseline stenosis resistance was calculated as the ratio of mean trans-stenotic pressure gradient to bAPV with a cut-off value of 0.66 [19], whereas hyperaemic stenosis resistance was calculated as the ratio of mean trans-stenotic pressure gradient to hAPV, with a cut-off value of 0.80 [20]. Binary iFR ≤ 0.89, FFR ≤ 0.8, and hyperaemic stenosis resistance ≥ 0.8 mm Hg/cm/s were considered abnormal. Normal CFC was defined as a CFR ≥ 2.8 and an hAPV of ≥ 49.0 cm/s. Mildly reduced CFC was defined as a CFR < 2.8 but > 2.1 and an hAPV of < 49.0 but > 33.0 cm/s. Moderately reduced CFC was defined as a CFR ≤ 2.1 and > 1.7, and an hAPV of ≤ 33.0 and > 26.0 cm/s. Finally, severely reduced CFC was defined as a CFR ≤ 1.7, and an hAPV of ≤ 26.0 cm/s [21]. Abnormal CFC was defined as a moderately to severely reduced CFC.

All analyses were performed at the lesion level, except for baseline patient characteristics. Continuous data were presented as mean ± standard deviation or median (first, third quartile [Q1, Q3]), and were compared by using the paired Kruskal-Wallis test. Analyses across iFR/FFR concordance and discordance groups were compared with 1‑way ANOVA, Kruskal–Wallis, χ² or Fisher’s exact test. Receiver operating characteristic (ROC) curves were computed to compare the diagnostic efficiency of each invasive physiological index against severely or moderately reduced CFC by the area under the ROC curve (ROC_AUC). ROC_AUC was calculated using DeLong’s method. Using the clinically established FFR and iFR cut-off values (≤ 0.80 for FFR and ≤ 0.89 for iFR), diagnostic agreement, sensitivity, specificity, positive predictive value, and negative predictive value were evaluated between FFR and iFR. Applicable tests were 2‑tailed and p < 0.05 was considered statistically significant. For all statistical analyses, the STATA version 15.1 (StataCorp, College Station, Texas) software package was used.

Patient characteristics of FFR/iFR discordant groups are shown in Tab. 3. FFR and iFR measurements were discordant in 15% (n = 97) cases, comprising of 57% (n = 55) lesions with FFR+/iFR− and 43% (n = 42) lesions with FFR−/iFR+ (Fig. 1). Patients in the FFR+/iFR− group were younger (p = 0.02) and were more frequently active smokers (p = 0.04), whereas patients with FFR−/iFR+ were more frequently diabetic (p = 0.08). For the left anterior descending artery (LAD), a total of n = 60 (15%) lesions (n = 30 [FFR+/iFR−] and n = 30 [FFR−/iFR+]) were discordant. For the right circumflex artery (RCX), a total of n = 26 (20%) lesions (n = 19 [FFR+/iFR−] and n = 7 [FFR−/iFR+]) were discordant. For the right coronary artery (RCA), a total of n = 11 (9%) lesions (n = 6 [FFR+/iFR−] and n = 5 [FFR−/iFR+]) were discordant. Thus, discordance occurred most frequently in the RCX.

Lesion and physiology characteristics across FFR/iFR groups are summarised in Tab. S1 in ESM. bAPV and hAPV were not significantly different between the discordant groups (p = 0.202 and p = 0.09 respectively) (Fig. 2a). CFR was significantly different across groups (p = 0.0001). Lesions with FFR+/iFR− discordance had similar hAPV and CFR compared with lesions with FFR−/iFR− concordance (hAPV 31 cm/s [Q1, Q3: 23, 44] and CFR 2.4 [Q1, Q3: 2.0, 2.7] versus hAPV 34 cm/s [Q1, Q3: 25, 44] and CFR 2.4 [Q1, Q3: 2.0, 2.9] respectively [p > 0.05 for all]). In contrast, lesions with FFR−/iFR+ discordance had similar hAPV and CFR compared with lesions with FFR+/iFR+ concordance (hAPV 29 cm/s [Q1, Q3: 19, 37] and CFR 1.6 [Q1, Q3: 1.4, 2.1] versus hAPV 24 cm/s [Q1, Q3: 17, 34] and CFR 1.6 [Q1, Q3: 1.3, 2.1] [p < 0.001 for all]) (Fig. 2b,c).

Figure 3 shows the distribution of lesions across the CFC categories within each of the FFR/iFR groups. Lesions with FFR+/iFR− discordance had abnormal CFC in 22% of cases, similar to lesions with FFR−/iFR− concordance where 21% of cases had abnormal CFC (p = 0.64). Lesions with FFR−/iFR+ discordance had abnormal CFC in 55% of cases, similar to lesions with FFR+/iFR+ concordance where 63% of lesions had abnormal CFC (p = 0.28).

In ROC analysis, iFR showed better diagnostic efficiency than FFR for the identification of abnormal CFC (ROC_AUC: 0.74 versus 0.68 respectively: p < 0.001) (see Fig. S1 in ESM).

Baseline stenosis resistance was not significantly different between the discordant groups (p = 0.093), but hyperaemic stenosis resistance was 0.70 (0.50, 0.96) for FFR+/iFR− lesions versus 0.42 (0.36, 0.63) for FFR−/iFR+ lesions (p < 0.001). Baseline stenosis resistance was highest for vessels with FFR−/iFR+ discordance and hyperaemic stenosis resistance was highest for vessels with FFR+/iFR− discordance. Moreover, there was a significant difference in CFR and delta APV across the FFR/iFR groups and across the discordance groups specifically: delta APV 18 (13, 24) and CFR 2.4 (2.0, 2.7) for the FFR+/iFR− group versus delta APV 11 (6, 17) and CFR 1.6 (1.4, 2.1) for the FFR−/iFR+ group (p < 0.001 for delta APV and p < 0.001 for CFR) (see Tab. S1 and Fig. S2 in ESM).

In vessels with FFR+/iFR− discordance, iFR values were close to the cut-off value (iFR 0.91 [Q1, Q3: 0.90, 0.93] versus FFR 0.75 [Q1, Q3: 0.73, 0.77]). Similarly, in vessels with FFR−/iFR+ discordance, iFR values were close to the cut-off value (iFR 0.87 [Q1, Q3: 0.84, 0.88] versus FFR 0.84 [Q1, Q3: 0.82, 0.86]) (Tab. S1 in ESM).

Vessels with FFR+/iFR− discordance had similar baseline stenosis resistance compared with vessels with FFR−/iFR+ discordance (0.45 mm Hg/cm/s [Q1, Q3: 0.31, 0.63] versus 0.59 mm Hg/cm/s [Q1, Q3: 0.35, 0.76] respectively; p = 0.093). In contrast, vessels with FFR+/iFR− discordance had higher hyperaemic stenosis resistance compared with vessels with FFR−/iFR+ discordance (hyperaemic stenosis resistance 0.70 mm Hg/cm/s [Q1, Q3: 0.50, 0.96] versus 0.42 mm Hg/cm/s [Q1, Q3: 0.36, 0.63] respectively; p < 0.001) (see Fig. S2 in ESM).

Similar to previous reports, discordance between FFR and iFR occurred in 15% of cases in the current analysis [6]. As depicted in Tab. S1 in ESM, the main difference between the discordance groups is reflected by the delta APV, CFR and hyperaemic stenosis resistance. This indicates the direct relationship of the discordance between FFR and iFR to the delta APV per interrogated vessel and the proportionally linear relationship between pressure and flow across a stenosis [22]. This relationship can be described by the quadratic equation of the form \(\Updelta \mathrm{P}=\mathrm{Av}+\mathrm{B}v^{2}\); where \(\Updelta \mathrm{P}\) is the pressure drop across the stenosis, v is the flow velocity and A and B are stenosis-specific coefficients. As such, the pressure drop is quadratically dependent of the flow through it and this could explain the differences between the discordance groups. In the FFR+/iFR− group, the intermediate lesions are characterised by a higher epicardial resistance (hyperaemic stenosis resistance) and increased CFR as a larger delta APV results in a larger pressure drop across that lesion during hyperaemia compared to the resting conditions, and thus, an abnormal FFR in the presence of a normal iFR. In comparison, within the FFR−/iFR+ group, the delta APV is limited and together with a lower hyperaemic stenosis resistance results in a limited pressure drop during hyperaemia compared with that at rest, resulting in a normal FFR in the presence of an abnormal iFR (Tab. S1 in ESM).

The present study confirms, and expands on, previous studies on flow characteristics in FFR/iFR discordance. Petraco et al. [23] and Cook et al. [5] were the first to provide insights in the origin of FFR/iFR discordance. The shared finding of these studies was that vessels with FFR+/iFR− have higher hyperaemic flow and CFR compared with FFR−/iFR+ vessels. Our findings provide further evidence for the role of maximal flow values in FFR/iFR discordance (Tab. S1 in ESM). Patients in the present study with FFR−/iFR+ discordance were significantly older than patients with FFR+/iFR− discordance (mean age 66 ± 10 versus 62 ± 10; p = 0.02) and tended to have a higher prevalence of diabetes (p = 0.08). Increasing age and diabetes are associated with a diminished response to a potent vasodilator, consequently leading to lower maximal flow values, impacting FFR/iFR discordance as described above [1, 24, 25]. Since non-hyperaemic stenosis physiology, assessed by baseline stenosis resistance, was similar across the two discordance patterns, the physiological origin of FFR/iFR discordance lies in hyperaemic vessel flow characteristics that are not related to stenosis severity. This is further supported by the distribution of CFC across the FFR/iFR groups (Fig. 3).

High hyperaemic flow and CFR are predictors of benign long-term clinical outcomes, even when FFR ≤ 0.80 [14]. The previously reported reports from the DEFINE-FLOW study and the ILIAS-registry indicated that patients with vessels with normal CFR and abnormal FFR in whom revascularisation was deferred, have outcomes similar to those patients who were treated with revascularisation [26]. Moreover, low CFR is independently associated with poor long-term clinical outcome [12, 27‐29]. Hence, iFR seems efficient in identifying those stenoses that are associated with impaired flow characteristics since the current data confirm in detail that FFR/iFR discordance occurs on the basis of variable maximal flow values that are generally more benign in iFR− vessels without differences in stenosis severity. In this population, this efficiency was not increased by additional measurement of FFR. These findings should trigger further evaluation of the prognostic impact of FFR/iFR discordance, on which evidence remains scarce [30]. From the comprehensive perspective of coronary haemodynamics provided by CFC, FFR/iFR discordance due to abnormal FFR is associated with normal or mildly reduced CFC in nearly 80% of cases. This may explain the lower revascularisation rates noted in the iFR arms of the DEFINE-FLAIR and iFR SWEDEHEART which, as noted above, were not associated with worse outcomes than the FFR-based strategy. Alternatively, FFR/iFR discordance due to normal FFR might look worrisome: around 50% of these cases have moderately or severely reduced CFC. The prognostic implications of this finding have been highlighted in previous works [21]. It remains unclear if revascularisation of stenosis included in this category would be associated with improved patient outcomes. Moreover, other causes of iFR/FFR discordance have been suggested: lesion location and severity, heart rate, age and beta blocker use affect mainly FFR and should be taken into account [31]. It was suggested that non-hyperaemic indices may be less reliable in proximal LAD lesions, since these supply a large amount of subtended myocardial mass [32]. However, this physiological consideration is based on the relationship between larger subtended myocardial mass and higher maximal hyperaemic flow across the stenosis, which leads to higher pressure gradients across a given stenosis. As discussed above, this phenomenon is indeed associated with discordance between iFR and FFR, but also with benign coronary flow characteristics. This is supported by clinical outcome data from a combined analysis of DEFINE-FLAIR and iFR-SWEDEHEART as well, documenting a lower incidence of adverse events for LAD lesions deferred based on iFR measurements compared with LAD lesions deferred based on FFR measurements.

In clinical practice, borderline iFR values will be followed by FFR measurements for clinical decision-making. In general, a borderline iFR value with an abnormal FFR value will be interpreted as an indication for percutaneous coronary intervention, while percutaneous coronary intervention will be postponed in case of a normal FFR. The present study, including the ROC analysis comparing iFR and FFR, indicates an opposite interpretation: a patient may benefit from an intervention in case of iFR+/FFR−, while conservative therapy may be justified in case of iFR−/FFR+. In case of doubt, performing CFR measurements may provide a more robust answer whether it is safe to defer a certain lesion of intermediate severity, since hyperaemic flow is significantly different between FFR+/iFR− and FFR−/iFR+ lesions, where FFR+/iFR− lesions are associated with higher CFR and thus benign long-term clinical outcomes.

The total number of discordant iFR/FFR lesions in the DEFINE-FLOW and IDEAL is small, but these two studies combined provide the largest multi-centre analysis of patients with intermediate coronary lesions undergoing invasive physiological interrogation by combined pressure and flow velocity measurements. Second, clinical follow-up after coronary physiological assessment was not routinely performed in IDEAL, prohibiting the evaluation of clinical outcomes.