This was a single-center registry of all patients who underwent coronary angiography with functional assessment of coronary lesions by FFR and/or iFR over a five-year period (2012-2016). Whether to perform FFR, iFR or both was left to the operator's discretion, as was the decision whether to proceed to revascularization or medical treatment. Patients enrolled in clinical trials involving these techniques were excluded, since they were subject to specific protocols that did not rely on the operator's decision. Demographic, clinical and procedural characteristics were recorded prospectively, including FFR and iFR measurements and their results. This study is a retrospective analysis, since it focused mainly on the subgroup who underwent both measurements, and the registry was not created specifically for the purpose of comparing the two groups.

FFR was measured after maximum hyperemia was obtained by central venous administration of adenosine (140 μg/kg/min), using St. Jude Medical (now Abbot) or Volcano (now Philips Volcano) systems. The latter (console and other equipment) was used in all cases when iFR was measured alone or jointly with FFR.

Receiver operating characteristic curve analysis was used to identify the iFR value with the best diagnostic accuracy, using FFR as a reference with cutoffs of ≤0.80 and <0.75. Bland-Altman plots and Pearson's correlation were used to compare iFR and FFR values.

iFR results were also compared according to the hybrid strategy,7 in which a value between 0.86 and 0.93 was considered inconclusive, <0.86 was considered positive (indicating ischemia and therefore PCI), and >0.93 was considered negative (indicating absence of ischemia, and hence deferral of the procedure). The same assessment was carried out using a cutoff of ≤0.89 in accordance with two recent randomized trials.9,12 Both analyses used a cutoff of ≤0.80 for FFR as a reference.

Finally, the effect of iFR on fluoroscopy time, radiation dose, procedure time and use of adenosine was analyzed.

The Student's t test was used to compare quantitative variables and the chi-square test was used to compare qualitative variables. Values are presented as mean ± standard deviation (SD).

Over a five-year period, 326 patients (mean age 67±11 years, 68.7% male) underwent functional assessment of moderate coronary lesions.

Measurements were taken for 402 lesions. FFR was used to assess 376 (93.5%) lesions and iFR to assess 180 (44.8%), with FFR only used in 222 lesion (55.2%) and iFR only in 26 lesion (6.5%) (Figure 1). There were no complications arising from adenosine use. Table 1 summarizes the main demographic and clinical characteristics of the patients in the overall population and according to the use of FFR, iFR or both techniques. There was a higher prevalence of dyslipidemia in the iFR-only group than in the FFR-only group, and a higher prevalence of previous history of myocardial infarction (almost reaching statistical significance, p=0.05) and previous PCI (p=0.034) in the FFR-only group vs. the group in which both techniques were used. There were no other significant differences between the groups.

Figure 1.

Distribution of lesions according to method of functional assessment used. FFR: fractional flow reserve; iFR: instantaneous wave-free ratio.

(0.04MB).

In cases of non-ST-elevation acute coronary syndrome, the measurements taken were of non-culprit lesions and were performed 48-72 hours after symptom onset, while in ST-elevation acute coronary syndrome, six non-culprit lesions were assessed, all during the acute phase. Four of these assessments used both techniques and two used FFR only, with the FFR result used for decision-making. Of the four iFR measurements, three lesions had iFR 0.9 (two cases with positive and one with negative FFR), and the fourth had iFR 0.88 (FFR 0.81).

Table 2 shows the characteristics of the lesions assessed.

Mean iFR was 0.90±0.1, and was >0.93 in 60 cases (33.3% of all measurements) and <0.86 in 21 cases (11.7% of all measurements). In the iFR-only subgroup, all values were <0.86 or >0.93, so calculating the mean and SD does not represent its true distribution (iFR-only was the only subgroup without a normal distribution); when both techniques were used, mean iFR was 0.9±0.05. Mean FFR in the overall population was 0.82±0.08. No statistically significant differences were found between mean FFR for FFR-only cases (0.81±0.09) and cases in which both techniques were used (0.82±0.07). Figure 2 shows the distribution of overall iFR and FFR values.

Figure 2.

Overall distribution of instantaneous wave-free ratio and fractional flow reserve measurements. FFR: fractional flow reserve; iFR: instantaneous wave-free ratio.

(0.12MB).

A moderately significant correlation was found between iFR and FFR (R=0.504; p<0.001) (Figure 3), as shown in the scatter plot. Bland-Altman analysis showed good concordance (the mean difference between iFR and FFR was 0.085±0.062) (Figure 4).

Figure 3.

Scatter plot showing correlation between instantaneous wave-free ratio (iFR) and fractional flow reserve (FFR) measurements.

(0.07MB).

Figure 4.

Concordance between instantaneous wave-free ratio (iFR) and fractional flow reserve (FFR) measurements by Bland-Altman analysis.

(0.09MB).

In terms of diagnostic accuracy, the area under the curve (AUC) of iFR was 0.73 (95% CI: 0.65-0.81) and 0.72 (95% CI: 0.60-0.84), using cutoffs of ≤0.80 and <0.75, respectively, for FFR as reference (Figure 5).

Figure 5.

Receiver operating characteristic (ROC) curve analysis of instantaneous wave-free ratio (iFR) measurements according to the two reference values of fractional flow reserve (FFR) (≤0.80 and <0.75). AUC: area under the curve; CI: confidence interval.

(0.2MB).

The optimal iFR cutoff was found to be ≤0.91 for both FFR ≤0.80 (55% sensitivity, 79% specificity, 52% positive predictive value, 81% negative predictive value) and FFR <0.75 (47% sensitivity, 89% specificity, 19% positive predictive value, 97% negative predictive value). In addition, for FFR ≤0.80, the iFR range with sensitivity and specificity >90% was 0.88-0.94, and for FFR ≤0.75 it was 0.87-0.95.

An iFR between 0.86 and 0.93 was strongly associated with the decision to perform FFR (chi-square=30.1; p<0.001), which occurred in 93 out of 94 cases. When iFR was not within this range, the decision to assess the lesion by FFR was made in 55 cases (69.1%). No differences were found in the decision to perform FFR in this subgroup between iFR results of <0.86 vs. >0.93 (71.4% vs. 68%; p=0.792). In these cases, a statistically significant concordance of 87.3% was found between iFR and FFR results (chi-square=22.43; p<0.001). However, there was discordance between the two methods in four out of 13 (30.8%) cases of positive iFR with negative FFR, and in three out of 42 (7.1%) cases with negative iFR and positive FFR. This difference was statistically significant (p=0.026) (Figure 6).

Figure 6.

Flowchart of concordant and discordant instantaneous wave-free ratio (iFR) and fractional flow reserve (FFR) measurements.

(0.08MB).

These discrepancies are detailed in Table 3. Four of them were in measurements of the left anterior descending artery (16% of a total of 25 measurements in this artery), two in the right coronary artery (14% of 14 measurements) and one in the circumflex artery (6.7% of 15 measurements). There was only one measurement using both techniques in the left main, with no discordance. As can be seen, in cases of negative iFR and positive FFR, the operators chose to follow the FFR indication when the measurement was clearly less than 0.80. In case 3, the presence of a borderline value led the operator to defer the procedure. In cases of positive iFR, the operators followed the indication given by FFR.

Using an iFR cutoff of 0.89, 56 lesions (36.4%) were positive (≤0.89) and 98 (73.6%) were negative (>0.89). A statistically significant concordance of 72.1% (p<0.001) was found between the two techniques, while no significant differences were found between cases of positive iFR with negative FFR (20 out of 43 cases; 46.5%) and negative iFR with positive FFR (23 out of 43 cases; 53.5%).

The decision whether to proceed with revascularization was taken by each operator individually for each patient.

In the 26 cases in which iFR only was performed, all values were outside the gray area (seven cases <0.86, all others >0.93), and the decision on revascularization was made according to the iFR result.

When FFR only was performed, the treatment decision was made according to the FFR result (cutoff 0.80) in 97.7% of cases. In the other 2.3%, the operator decided not to act on the FFR value due to other relevant data, such as lesion characteristics, symptoms, or results of non-invasive ischemia tests.

When both techniques were performed, the operators followed the results whenever there was concordance between the two techniques. Cases in which there was discordance are described above (Table 3).

No statistically significant differences were found in fluoroscopy time, radiation dose or procedure time between patients who underwent iFR only and those who underwent FFR only, or between those who underwent both iFR and FFR and those who underwent only one technique (Table 4).

Considering only patients who underwent iFR only or both iFR and FFR (a total of 147), 55 (37.4%) of them had all lesions assessed and had a conclusive iFR measurement (outside of the 0.86-0.93 range). With the hybrid strategy, the use of adenosine in these patients would have been avoided.

In this registry, iFR showed reasonable diagnostic accuracy and concordance with FFR, although there were some differences from previous publications. The best cutoff in this sample was slightly higher than that published in other studies (0.91 vs. 0.897,13), regardless of the FFR value used as standard. In addition, its diagnostic accuracy was inferior to that of some other studies, in which the AUC ranged from 0.86 to 0.90,7,13,14 but better than in the VERIFY study.15 The correlation between iFR and FFR was also lower than in previous studies (0.504, compared to 0.81 in ADVISE II7). These results may reflect a sample-size effect, since our population was smaller than in the other studies, or selection bias, since the study was based on a registry. Operators may have chosen to use both measurements in particularly borderline cases, as can be seen by the fact that the mean iFR value was very similar to the diagnostic value used in DEFINE-FLAIR9 and iFR-SWEDEHEART12 (0.89). Furthermore, when iFR was performed first and the result was then confirmed using FFR, not only was the value borderline, the SD was also very small (0.05).

When the hybrid strategy was used, iFR showed high concordance within the diagnostic range (<0.86 and >0.93). This is similar to the results of the study by Escaned et al. that properly classified 91.6% of stenoses,7 which was slightly higher than in our case (87%). When a cutoff of 0.89 was used, the degree of concordance was also lower than in Escaned et al. (72.1% vs. 82.5%7), because this cutoff did not have the highest diagnostic accuracy in this sample (which limits the extent to which this result can be generalized), as well as the above-mentioned limitations.

Notwithstanding the good results from the hybrid approach in this registry, the degree of discrepancy between the two techniques was not negligible. When iFR and FFR gave different indications, operators tended to follow current international guidelines1 and use FFR as the gold standard. Discordance predominantly involved positive iFR and negative FFR, as in other studies.7 Thus, it may particularly important for the operator to proceed with FFR in borderline cases, especially when iFR is positive.

The recently published results of DEFINE-FLAIR9 and iFR-SWEDEHEART10 have for the first time provided solid evidence in favor of decision-making based only on iFR ≤0.89. At the same time, despite the unfavorable results of the FUTURE trial,16 the experience gained with FFR and the extensive literature on this technique cannot be ignored. Thus, if there are to be large multicenter randomized trials on the use of both techniques, two key issues must be addressed: what to do if the FFR and iFR results do not agree, and whether only one of the techniques should be used and, if so, which.

In the absence of agreement on the answers to these questions, certain points should be taken into consideration in order to arrive at the best decision in each case.

Firstly, the operator must ensure that the technique is performed correctly, with appropriate caliber, type and positioning of the catheter, to avoid attenuation or distortion of the pressure curves, and that pressures are correctly calibrated to ensure that the curves overlap in time and in quantity, synchronized with the electrocardiogram. Although both techniques are influenced by these factors, faults in execution may distort one measurement more or less than the other. For example, in iFR, failure to synchronize with the electrocardiogram can mean that the timing of the distal coronary pressure to aortic pressure ratio (Pd/Pa) reading is inaccurate, leading to non-diastolic or partially diastolic values being incorporated into the result. Another problem to consider is pressure drift; to correct this, the operator should confirm that the pressure curves overlap before ending the procedure and, if necessary, reposition the pressure wire proximally. Finally, it should be noted that the only iFR systems currently available are from Volcano Philips, and so all iFR measurements in our study were performed using this equipment; any measurement taken using other systems remains off-label. It should borne in mind that in our registry, the centers and all its operators were thoroughly experienced in the use of FFR, so we can assume that the error rate was low. Nevertheless, the possibility of inaccuracies in iFR measurements cannot be completely excluded, especially while the new technique was being introduced. Furthermore, although confirmation of the absence of pressure drift is routine practice at our site, no record was kept of this. Similarly, although measurements in which the electrocardiographic tracing was suboptimal were rejected and iFR was rarely used in patients with atrial fibrillation or flutter (in none of whom discordance between the two techniques was observed), the possibility that cardiac rhythm influenced some of the results cannot be completely ruled out.

Secondly, even if there are no technical failures, the physiology of what is being measured is relevant. Since FFR is dependent on maximal hyperemia, response can be affected by the drug used (some centers use vasodilators other than adenosine) and its administration route,17 the vasodilatory effect of the contrast itself,18 the presence of concomitant microcirculatory dysfunction,19,20 and adenosine's systemic vasodilatory effect. The response to the drug, at the level of both small and large coronary arteries as well as systemically, is also known to vary between individuals, especially in acute coronary syndrome.

Finally, it should be noted that steady-state FFR often does not match the nadir value, and so the former should be used,21 as in this registry. In our study, there were only five cases of discordance when the hybrid strategy was used in stable angina and two in acute coronary syndrome (both unstable angina), and so the clinical context likely had little effect on discordance.

The limitations governing the use of iFR should also be taken into account. Although iFR is not considered to be limited by hyperemia, Berry et al. dispute this and also claim that its diagnostic accuracy is lower than observed in other studies.15 In addition, recent randomized trials had few events and were non-inferiority studies in which the non-inferiority margin was relatively wide, as the authors acknowledged.9,10

Some authors suggest that the discrepancy between the two methods is greater in lesions involving the anterior descending artery,22 especially in its proximal segment.23 Given the low number of discordant findings in each vessel in our registry, it is not possible to draw conclusions about the effect of the vessels or segments involved. Even in the anterior descending, there were only four cases of discrepancy, a similar rate to that in the right coronary.

The results from our registry, together with findings in the literature, thus suggest that iFR and FFR are in fact complementary methods for the assessment of coronary lesions, and are not mutually exclusive. Initial assessment by iFR only appears to be the most sensible strategy, given its simplicity and speed of execution. However, the limitations of the above-mentioned studies suggest that caution is in order and that in doubtful cases operators should consider FFR as well. Concordance between the two values is a strong indicator when deciding whether to proceed with revascularization. Understanding the cause of any discrepancy is crucial in order to decide if one of the results is more valid for an individual lesion and patient. If the cause is unclear, the additional use of invasive and/or non-invasive imaging methods should be considered before reaching a final decision, especially as iFR and FFR have mainly been validated for stable coronary artery disease.

In addition to not requiring adenosine, one of the theoretical advantages of iFR is that it may reduce procedure time. However, there is little evidence in the literature to support this. Although the DEFINE-FLAIR study9 reported a significant reduction of 4 min 30 s min with iFR compared to FFR, the reduction was not significant in the iFR-SWEDEHEART study.10 As in the latter study, in our registry the technique did not affect this or other parameters (radiation dose and fluoroscopy time). This may be explained by the significant differences between the patients studied, with the operator opting for a simpler technique in more severe cases. Nevertheless, there were few statistically significant differences between the groups: absolute and relative differences were small and had little statistical significance or clinical relevance, and appeared to be unrelated to the operators’ decisions. In addition, few patients underwent iFR only, which also limits comparisons between groups. Furthermore, in each group, in some cases more than one lesion was assessed. This may have contributed to the numerical differences between the groups, although these were not statistically significant.

Finally, the apparent discrepancy (not statistically significant) between radiation dose and procedure time (longer procedure time but lower radiation dose in the iFR group) resulted from the use of different frame rates by different operators, so the radiation dose did not always correspond to the fluoroscopy duration. Between groups, it was found that longer fluoroscopy duration was not associated with increased radiation dose. This is why we opted to present data on both groups.

In this registry, iFR was used at the operator's discretion, and especially in the early stages of its implementation, both techniques were commonly used, which resulted in a high rate of adenosine use. As pointed out in the Results section, if the hybrid strategy had been followed, the use of adenosine could have been avoided in almost 40% of patients. This figure is still lower than in other studies, particularly the ADVISE II study,7 in which the hybrid strategy led to 65.1% of patients not being administered adenosine. The difference between these figures is related to the observational nature of our study, in which there were no set criteria for selecting techniques, and which could have led to selection bias in cases with borderline iFR values.