This is a prospective open label study including patients with permanent or persistent AF and indication of coronary angiography (suspected chronic coronary syndrome based on clinical evaluation and non-invasive coronary imaging, or suspected acute coronary syndrome) showing intermediate coronary stenosis (40%–70% visually-assessed stenosis diameter) requiring routine physiological assessment (n = 18 vessels from 14 patients). The AF was established to be permanent or persistent in accordance with the 2020 ESC guidelines on AF (12). Patients presented either permanent AF (n = 8 patients)—a therapeutic attitude consisting of no additional attempts to restore sinus rhythm, or persistent AF (n = 6 patients)—patients who were cardioverted (drugs or electrical cardioversion) usually by the end of the same hospitalization. The exclusion criteria were: cardiac arrest, cardiogenic shock, acute decompensated heart failure, acute phase of ST segment elevation myocardial infarction, culprit vessel in ST segment elevation myocardial infarction and non-ST-segment elevation myocardial infarction, contraindications to adenosine administration (acute asthma, high degree atrioventricular block), and patients in sinus rhythm during the procedure. The prospective study was approved by the Ethical Committee of CHU Brugmann (reference number CE 2020/17). Informed consent was obtained from all patients undergoing investigation.

We measured FFR (fractional flow reserve), and Doppler-derived coronary flow indices—CFR (coronary flow reserve), BSR (basal stenosis resistance), HSR (hyperaemic stenosis resistance), BMR (basal microvascular resistance), and HMR (hyperaemic microvascular resistance) using a ComboWire® (Philips, Volcano) guidewire. Hemodynamic measurements were performed under basal conditions and under hyperaemia induced by intracoronary administration of adenosine (dose of 150 μg, increased progressively until attaining maximal hyperaemia). CFR is calculated as the ratio of maximum blood flow during hyperaemia (hAPV—hyperaemic average peak velocities) and resting coronary blood flow (bAPV—baseline average peak velocities), and reflects both the epicardial and the microcirculatory coronary status (13). Given the irregular RR interval in AF with potential impact on coronary flow, APV was calculated as the average instantaneous peak velocity over an interval of 5 beats. HSR is a combined pressure- and flow-based index defined as the ratio of hyperaemic stenosis pressure gradient (Pa-Pd) and hAPV. A previous study has shown superior diagnostic accuracy of HSR compared to FFR or CFR for assessing the functional significance of epicardial coronary lesions in the general population (13), although is not currently implemented in day-to-day practice. Analogous to HSR, BSR is a combined pressure- and flow-based index that evaluates the epicardial coronary stenosis under basal conditions, being defined as the ratio of resting pressure gradient (Pa-Pd) and bAPV. The microvascular coronary resistances are reflected by BMR under resting conditions (resting Pd/bAPV), and by HMR after adenosine administration (hyperaemic Pd/hAPV) (14).

We assessed the agreement of CFR and FFR by evaluating the concordance of lesion classification obtained using the two techniques. Criteria for revascularization were FFR ≤ 0.8 at maximal hyperaemia (15), and a CFR < 2.5 was considered indicative of coronary microvascular dysfunction (CMD) (16, 17). In addition, we compared FFR to HSR values. HSR higher than 0.8 mmHg/cm/s is associated with significant epicardial coronary artery disease (18).

Patients with AF and non-obstructive epicardial coronary disease (FFR > 0.8; 15 vessels from 13 patients) were divided in two subgroups based on the presence of increased microvascular resistance [HMR ≥ 2.5 mmHg/cm/s (19, 20)]. We compared patient and vessel characteristics between the two subgroups (HMR < 2.5 mmHg/cm/s vs. HMR ≥ 2.5 mmHg/cm/s), including differences in pressure- and flow-based indices.

In patients with non-obstructive epicardial coronary disease, the coronary physiology indices were compared based on the presence of permanent or persistent AF. In addition, the presence of persistent symptoms was evaluated at 3 months follow-up by the treating cardiologist. The symptoms motivating the coronary angiography were either chest pain or dyspnoea. A systematic differential diagnostic work-up performed prior to the coronary angiography (extensive blood work-up, chest imaging, echocardiography) excluded other potential causes for the presence of the above specified symptoms. Persistent symptoms were defined as the continuous presence of chest pain or dyspnoea at 3 months follow-up. The treating cardiologist was aware of the FFR values, but not of the values of flow-derived coronary indices, at the moment of clinical assessment.

Statistical analyses were performed using GraphPad Prism 9.0.1. Data are presented as mean ± standard deviation. For analysing two groups we employed the Mann-Whitney U test. Categorical variables were compared using Chi-square test. Correlation analyses were performed using Spearman's correlation test. The maximum hyperaemic response was analysed using two-way ANOVA, followed by Tukey post-hoc test. Significance level was set at 0.05.

We performed combined assessment of coronary pressure- and flow-derived indices of a total of 18 vessels from 14 patients with AF. The demographic and clinical characteristics of these patients are summarised in Table 1, and of the corresponding vessels in Table 2.

From the analysed vessels, 18/18 vessels (100%) presented a pathological CFR (<2.5), and only 3/18 (17%) demonstrated obstructive epicardial coronary disease (FFR ≤ 0.8) requiring percutaneous coronary angioplasty. These 3 pathological vessels were distributed in 3 different patients, 2 of which showed one vessel with pathological FFR, and one vessel with non-pathological FFR. As such, most vessels (15/18; 83%) showed discordant FFR/CFR with preserved FFR and low CFR, while 3/18 (17%) showed concordant FFR/CFR with low FFR and low CFR (Figure 1A). We detected a moderate correlation between FFR and CFR (r = 0.5840, p = 0.0109; Spearman correlation) (Figure 1B).

Vessels of patients with AF and non-obstructive epicardial disease (FFR > 0.8) presented low CFR values (mean value of 1.74 ± 0.24), indicative of CMD. There was no correlation between the CFR values and the left ventricular ejection fraction, or the cardiac index assessed by transthoracic echocardiography (p > 0.05; Spearman correlation) as to explain the observed low CFR values. Using the described cut-off of 2.5 mmHg/cm/s for pathological HMR (19, 20), and the pattern of structural/functional CMD identified in the general population (16), we divided the vessels of patients with AF and non-obstructive epicardial disease as presenting structural CMD (HMR ≥ 2.5 mmHg/cm/s; 7 coronary arteries and 6 patients) or functional CMD (HMR < 2.5 mmHg/cm/s; 8 coronary arteries and 7 patients) (Tables 3, 4). The subgroup of vessels with structural CMD demonstrated significantly higher BMR values compared to the vessels with functional CMD (4.34 ± 0.67 mmHg/cm/s vs. 2.71 ± 1.43 mmHg/cm/s; p = 0.0289) (Table 4). High HMR was associated with a reduced microvascular response to adenosine defined as 100-(HMR/BMR) (%) when compared to low HMR (13.6 ± 11.4% vs. 37.5 ± 16.7%; p = 0.0093) (Figure 2A). In addition, there was a moderate inverse correlation between HMR values and the microvascular response to adenosine (Figure 2B) (r = −0.6429, p = 0.0116; Spearman correlation). In line with these findings, the maximum hyperaemic effect in vessels of patients with non-obstructive epicardial disease was found to depend on the presence of increased microvascular resistance, as adenosine administration led to a significant decrease in Pd/Pa only in vessels with HMR < 2.5 mmHg/cm/s (Figure 2C). A more detailed look at the clinical and lesion characteristics of patients with non-obstructive epicardial disease according to the presence of increased microvascular resistance is overviewed in Tables 3, 4. The subgroup with functional CMD presented higher bAPV (36.4 ± 23.4 cm/s vs. 17.4 ± 4.9 cm/s; p = 0.0294) and hAPV (62.4 ± 34.8 cm/s vs. 29.8 ± 10.7 cm/s; p = 0.0347) compared to the subgroup with structural CMD.

Patients with structural CMD demonstrated significantly more persistent symptoms (chest pain or dyspnoea) at 3 months following invasive assessment (83.3% vs. 0.00%, p = 0.0021) compared to functional CMD (Figure 3). Furthermore, permanent AF was more often associated with high HMR (3.13 ± 1.20 mmHg/cm/s vs. 1.46 ± 0.79 mmHg/cm/s, p = 0.0193) (Figure 4A) and high BMR (4.16 ± 0.83 mmHg/cm/s vs. 2.10 ± 1.26 mmHg/cm/s, p = 0.0127) (Figure 4B) compared to persistent AF.

Regarding the functional evaluation of coronary stenosis severity, there was high agreement in the classification of the lesions between FFR and HSR (83%), with the two indices showing strong correlation across all patients with AF (r = −0.7446, p = 0.0004; Spearman correlation) (Figure 5A). Moreover, FFR values correlated strongly with the microvascular component of total vascular resistance, defined as HMR/(HMR + HSR) (r = 0.9866, p < 0.0001; Spearman correlation) (Figure 5B).

All patients with AF in our study presented with diminished CFR (<2.5), even in the absence of obstructive coronary artery disease, indicative of CMD. There are multiple studies showing that myocardial ischemia follows more closely changes in coronary blood flow than coronary perfusion pressure (21, 22). Furthermore, low CFR values (<2.5), as observed in the current study, were associated with an excess of major adverse cardiovascular events (MACE) and target vessel failure at 5-year follow-up (17). Our findings of abnormal CFR in patients with AF are consistent with previous reports of decreased hyperaemic myocardial blood flow in AF assessed by non-invasive methods (6–9). In addition, Kochiadakis et al. demonstrated diminished CFR in acute experimentally-induced AF in subjects with normal angiographic coronary arteries by means of invasive coronary physiology assessment (10). This has been postulated to occur due to the irregular ventricular rhythm associated with AF. It took close to twenty years after this initial investigation for a more recent study to revisit the evaluation of CFR in AF using invasive coronary assessment. In this study, Ozcan et al. showed lower CFR values in patients with AF and non-obstructive coronary artery disease, hypothesised to second the presence of microvascular dysfunction, however without evaluating specifically the microcirculation (11). The observed low CFR in these patients could be partly due to the highly prevalent association of AF with heart failure (23), or it could be a purveyor of arrhythmia-induced cardiomyopathy. Furthermore, we observed a mean value of CFR of 1.74 in patients with AF and non-obstructive coronary artery disease, which is similar to that observed by Ozcan et al. (1.70) (11), however lower than that obtained by Kochiadakis et al. (2.15) (10). Although all three studies suggest the presence of CMD in AF as assessed by CFR (<2.5), the higher mean values obtained in the latter study could be explained by CFR measurements performed in acutely-induced AF (10).

Interestingly, we could observe a highly prevalent discordance between CFR and FFR (FFR > 0.8 and CFR < 2.5) in 83% of patients. In the general population, this particular discordance of FFR/CFR was observed in approximately 25% of patients according to recent studies (24, 25), although a previous study reported this discordant population in as low as 6.4% of cases (26). Patients with discordant FFR/CFR presented higher MACE rates than the concordant normal FFR/CFR group at 1 year (27) and 10 years (26) follow-up. Nonetheless, regardless of the high rate of discordance of FFR/CFR in terms of their established cut-offs, we could demonstrate a significant, albeit moderate correlation between the absolute FFR and CFR values in patients with AF, similar as reported in the general population (28).

HSR was found to have superior diagnostic accuracy compared to FFR and CFR when confronted to non-invasive imaging using single photon emission computed tomography (18), and has been used for evaluating the accuracy of pressure-based indices in the general population (29, 30). In the current study, we observed that 83% of patients showed a concordance of FFR and HSR, using a cut-off for HSR of 0.8 mmHg/cm (18). This is consistent with the previously reported diagnostic agreement between FFR and HSR in the general population of 78.7% (30). However, HSR is also dependent on an adequate hyperaemic response, and can be confounded by the presence of microvascular dysfunction (13). Previously, Kochiadakis et al. indicated that irregularities in ventricular rhythm may be responsible of diminished CFR in AF (10), while a recent case report by Mills et al. highlighted that changes in cardiac rhythm associated with AF may impact coronary physiological assessment (31). In a retrospective setting, we observed that FFR was more robust than instantaneous wave-free ratio (iFR), a resting coronary pressure index, in evaluating coronary lesions in AF, showing similar variability, reproducibility, and lesion classification as observed in patients with sinus rhythm (32). In addition, adenosine-induced hyperaemia during FFR measurement was found to be equivalent in AF compared to sinus rhythm in terms of magnitude and duration (33). On the other hand, iFR presented poorer reproducibility in lesion classification in AF, and was prone to increased beat-to-beat variability which correlated with the heart rate variability (32), supporting the notion that abnormalities in cardiac rhythm may impact evaluation of coronary physiology using resting physiological indices. Further studies are necessary to provide a better understanding on the impact of cardiac rhythm on coronary physiological assessment in AF.

The low CFR observed in patients with AF and non-obstructive epicardial coronary disease indicates the presence of CMD in this population (34). CMD is a result of functional or structural alterations of the endothelium (35). Mechanistically several pathogenic hypotheses were described including impaired NO-mediated vasodilatation, as a consequence of glycocalyx degradation due to oxidative stress, inflammation or ischemia (35), triggers which have been described in AF (36). In addition, CMD was also hypothesized to occur secondary to alterations in coronary flow distribution due to compromised intercellular signal transmission, or to impaired metabolic signalling (35). Furthermore, structural modification of the microcirculation such as capillary rarefaction and remodelling can contribute to increased coronary microvascular resistance (37). Previous studies have indicated the presence of microvascular coronary dysfunction in AF (11). Using positron emission tomography, Range et al. identified increased coronary vascular resistance under hyperaemic conditions in patients with AF (9). Similarly, Kochiadakis et al. observed increased hyperaemic coronary vascular resistance in subjects with acute experimentally-induced AF compared to sinus rhythm using invasive coronary assessment (10). The microvascular dysfunction observed in AF could be responsible for the ischemic manifestation (chest pain, increased troponin, ECG modifications) in the absence of significant epicardial coronary disease (38). In addition, the presence of microvascular disease in the general population was shown to be associated with poorer prognosis in terms of cardiac mortality (39).

The microvascular coronary status is reflected in terms of coronary indices by HMR and BMR. Recent studies have identified that HMR values above 2.5 mmHg/cm/s are suggestive of increased microvascular resistance (19, 20) defining structural CMD, whereas functional CMD is characterized by HMR < 2.5 mmHg/cm/s in presence of diminished CFR (16, 40). Interestingly, we identified amongst the population with AF and non-obstructive epicardial coronary disease, 47% (7/15) vessels showing HMR ≥ 2.5 mmHg/cm/s, which contrasts to approximately 21% reported in the general population (16). Using this threshold, we divided the patients with AF and non-obstructive epicardial disease as presenting structural CMD (HMR ≥ 2.5 mmHg/cm/s) or functional CMD (HMR < 2.5 mmHg/cm/s), demonstrating different physiological and clinical characteristics. Coronary arteries from patients with AF and structural CMD showed higher BMR values, impaired response to adenosine, lower bAPV and hAPV, and were more frequently associated with permanent AF.

High HMR values were described as a predictor of myocardial ischemia even in the absence of significant epicardial coronary lesions (14). In addition, high HMR values could influence the evaluation of coronary physiology using pressure-based indices (41, 42). Specifically in the AF population, a retrospective study found that iFR was less reproducible in assessing intermediate coronary stenoses (32), that could potentially be explained by the increased microvascular resistance. In addition, in the current study we have observed a strong correlation between FFR and the ratio between HMR and the total vascular resistance (HMR + HSR). Previously, we have reported similar hyperaemic responses to adenosine in an overall retrospective population of patients with AF compared to sinus rhythm, however without having available their microvascular resistance characteristics (33). In the present study, we showed that vessels from patients with AF and high HMR have a reduced hyperaemic response compared to the subgroup of AF and low HMR. Furthermore, we demonstrated that HMR values were inversely correlated with the magnitude of the response to adenosine. This aspect should be considered when assessing epicardial coronary lesion severity using FFR in patients with AF and structural CMD.

Patients with AF and functional CMD present significantly higher bAPV, which could explain the low CFR in this subgroup even in the presence of normal microvascular resistance, as there is a maximum reachable hyperaemia (43). Previous observations indicate that acute experimentally-induced AF leads to increases in baseline coronary blood flow, that could be partly explained by an increase in heart rate, and hypothesised to be the result of coronary vasodilation secondary to higher metabolic demands in AF (10). This is in line with our findings that patients with higher bAPV present lower BMR. Another potential explanation of high bAPV reported in the literature was the dysfunction of coronary autoregulation which was associated with adverse long-term outcomes (39). With increasing AF burden, endothelial dysfunction, sympathetic activation, progressive left ventricular fibrosis and both atrial and ventricular remodelling were described, that could impact irreversibly the coronary microcirculation (36). Accordingly, patients with AF and high HMR showed increased BMR and impaired response to adenosine, potentially preventing them from reaching maximum hyperaemia.

In our study at 3 months follow-up, patients with AF and non-obstructive epicardial coronary disease presenting with structural CMD demonstrated higher percentage of persistent symptoms, compared to patients with functional CMD. It should be noted that persistent symptoms were solely defined as present or absent, without providing specific quantification of symptom burden. It has been previously hypothesised that the increased microvascular resistance could be responsible for the symptoms suggestive of myocardial ischemia in AF even in the absence of significant coronary artery disease (36). Our findings seem to support this hypothesis, as we observed more persistent symptoms in patients with AF and increased microvascular resistance in the absence of obstructive coronary artery disease as assessed by FFR. Nevertheless, given the higher prevalence of persistent symptoms in the subpopulation of AF with increased HMR, and the strong correlation between FFR and the microvascular component of the total vascular resistance, possibly a part of the FFR/CFR discordance observed in patients with AF is related to an underestimation of coronary artery stenosis assessed by FFR. Indeed, in our study, solely 17% (3/18) vessels were found to have pathologic FFR (≤0.8), and thus requiring revascularization, which is in contrast with the general population estimates of about 45%–50% (44, 45). Future studies will be important to address this hypothesis. Interestingly, performing coronary angioplasty in cases of FFR negative lesions, but showing high HMR, has been described in literature as an efficient rescue therapy in case of persistent symptoms (42). Alternatively, the difference in clinical outcome may also be explained by the presence of predominantly permanent AF in the subgroup with structural CMD, and more prevalent persistent AF in the subgroup with functional CMD. Notably, only 2 out of the 7 patients with functional CMD and absence of persistent symptoms were in sinus rhythm at 3 months clinical follow-up.

Furthermore, as indicated above, it is the subpopulation of permanent AF which presents increased HMR and BMR compared to persistent AF. As such, particular attention is warranted clinically in interpreting the coronary pressure-based indices in this subgroup. Interestingly, in this study, 7 out of the 8 patients with permanent AF experienced AF for a duration of at least 1 year, while 4 out of the 6 patients with persistent AF presented de novo AF. This difference in the duration of AF between the two groups could perhaps impact the observed changes in microcirculation characteristics. This could mirror the observed bimodal pattern of coronary microvascular dysfunction (functional CMD, followed by structural CMD) described in diabetes mellitus in a time-dependent manner (40), and more recently in the general population (16).

We acknowledge that the present study has limitations, including the small sample size, especially regarding the clinical findings, driven by the current shortage of commercial ComboWire® guidewires, the lack of a control group in sinus rhythm, and the limited clinical follow-up, as it was designed firstly as a physiological single-centre study. In addition, the clinical evaluation included only the presence or absence of persistent symptoms, without providing specific quantification of symptom burden. Furthermore, our observations cannot be extrapolated to patients with paroxysmal AF, as they were not included in the study population.