The present study was a prospective, non-randomized trial with an open-label design. The non-randomized design was chosen since the patients constitute their own controls. The lower dose was always administered first in order to mimic clinical routine. The study was conducted as a single-center, non-consecutive trial at Skane University Hospital, Lund, Sweden. Patient admitted for coronary angiography were considered eligible for inclusion if the following criteria were fulfilled: Age ≥18 years, borderline-significant coronary stenosis (indication for FFR according to ESC Guidelines) [6] and signed informed consent prior to enrollment. Key exclusion criteria were allergy to adenosine or contrast media, baseline mean arterial pressure <60 mmHg, baseline heart rate <50, pharmacological treated asthma, chronic obstructive pulmonary disease equivalent to GOLD classification III and IV [38], confusion or inability to comprehend the study information.

Following coronary angiography and intracoronary administration of 200 μg Nitroglycerin, a 0.014-inch pressure guide wire (Primewire Prestige®/Verrata® Pressure Guide Wire, Volcano Corporation, San Diego, CA, US) was advanced through a 6-F guide catheter into the coronary artery, calibrated and subsequently advanced distal of the lesion. The infusion of intravenous adenosine (Adenosin Life Medical 5 mg/ml, Life Medical Sweden AB) was started at a weight-adjusted rate, equivalent to standard dose 140 μg/kg/min and terminated when the two minutes measurement was completed. The agent was administrated through a peripheral intravenous line. FFR was recorded for two minutes (±5 s) and calculated by the Volcano CORE™ integrated system with the S5I® software and Case Manager (Volcano Corporation, San Diego, CA, US). Prior to the second measurement, a recovery time was mandatory for the pressure curve to return to baseline values (minimum 5 min). After recovery, the second measurement was performed with similar FFR technique and an intravenous adenosine infusion of 220 μg/kg/min. FFR was considered significant if ≤0.80. The FFR results of standard dose were used for clinical decision of revascularization. A ≥0.02 drift of the FFR-wire was considered clinical relevant, and if this occured, a new calibration was performed. Consumption of caffeine was defined as a minimum of 200 ml filter coffee consumed ≤6 h prior to FFR. The patients coffee intake ranged between 200 and 400 ml.

Mean arterial pressure (MAP) and heart rate were monitored and collected at baseline and every 30 s during the measurements (Philips Intellivue® Cardiac system, version M8010A, Royal Philips Electronics, Amsterdam, Netherlands). The patients were asked to score maximum discomfort following each measurement by a validated tool for determination of pain/discomfort, the Visual-Analogue-Scale (VAS) [39‐41].

Based on standard deviations in FFR-measurements in previous studies [24, 25], a sample size of 72 would have a 90% power to detect a 15% difference of FFR with a significance level (alpha) of 0.05% (two-tailed). In order to compensate for loss of data due to a presumed inability to tolerate high dose adenosine, the a priori number of patients intended to include was 85. The statistical analysis was performed using the Graphpad Prism6™ for Mac OS X™ (Graphpad Software Inc. La Jolla, CA, US). Continuous variables were followed and reported as a mean ± standard deviation or as median (interquartile range 25–75) if asymmetrically distributed. The correlation of adenosine doses was calculated by Wilcoxon matched-pairs signed rank test and a linear regression model. The agreement was graphically displayed in a Bland-Altman plot. In addition, for the continuous hemodynamic variables of mean arterial pressure and heart rate, an area under curve (AUC) analysis was performed. The interference between the groups in the caffeine analyses was based on the computation of the Mann-Whitney U-test (Wilcoxon rank-sum-test) to determine significance. A p-value of <0.05 was considered statistically significant.

Eighty-seven patients scheduled for FFR were included. Ten patients (11%) developed atrioventricular block during administration of standard dose adenosine and were excluded from the second measurement and further analysis. Another 2 patients declined to participate in the second measurement due to severe discomfort from adenosine administration. Of the remaining 75 patients, two complete FFR measurements within the same coronary lesion using the two different dose regimens were performed. Patient demographics and clinical characteristic are presented in Table 1. In 28% of the cases, FFR was ≤0.80 after the first measurement and PCI were thus performed. In the group treated by PCI, procedural success was 100%. Stable angina (Canadian Class I-II) was the most common indication for FFR (36%). In the group of acute coronary syndrome, the non-culprit vessel was used for the current trial. In 49.3% of the cases, LAD was the target FFR vessel and in 41.3%, the lesion location was proximal (Table 2.) 89.5% of the patients were treated by dual antiplatelet therapy (Table 3).

There was no significant difference in the matched-pairs comparison of intravenous adenosine infusion of 140 μg/kg/min versus 220 μg/kg/min (0.85 [0.79–0.90] vs 0.85 [0.79–0.89], p = 0.24) (Fig. 1). In addition, high dose adenosine showed a strong significant linear correlation to standard dose (r = 0.86, slope = 0.89, p = <0.001) (Fig. 2). In the Bland-Altman analysis, average of the differences were -0.005 ± 0.03 (mean bias ± SD) [-0.07 to 0.06], [95% CI], (Fig. 3). In four patients (5.3%), the higher dose of Adenosine caused a change in agreement due to lowering FFR below the treatment threshold of 0.80 (0.85–0.79, 0.81–0.78, 0.81–0.79 and 0.81–0.79). The high dose did not decrease FFR below 0.75 in any of the 75 cases. Thus, all changes remained in the borderline region.

Mean arterial pressure and heart rate were similar during infusion of the different adenosine doses. MAP: Standard dose 7152 ± 178.2 versus high dose 6991 ± 346.7 AUC [arbitrary units], p = 0.34). Heart rate: Standard dose 5488 ± 95.45 versus high dose 5602 ± 49.10 AUC [arbitrary units] p = 0.11 (Fig. 4). Ten patients were excluded from the second FFR measurement due to atrioventricular block. In the remaining 75 patients, the occurrence of atrioventricular block and bradyarrhytmias was 5.3%. Patient maximal discomfort during adenosine administration, measured by VAS, was significantly higher in the dosage of 220 μg/kg/min (8.0 [5.0–9.0]) versus standard dose (5.0 [2.0–7.0]), p = <0.001 (Fig. 5).

A subgroup analysis was performed in the 43 patients (57%) of the study population who reported caffeine consumption ≤6h prior to FFR. In four patients (5.3%), three from the caffeine group (4%) and one from the control group (1.3%), high dose adenosine decreased FFR from non-significant to borderline significant (0.78–0.79). In the adenosine dosage of 140 μg/kg/min, FFR was significantly higher in the caffeine group compared to control (0.90 [0.83–0.93] versus 0.82 [0.75–0.85], p = <0.001. In the high dose regime, there was a similar trend but not significant (0.87 [0.81-0.91 versus 0.83 [0.77–0.88], p = 0.09) (Fig. 6). In a paired comparison of caffeine consumers in the study population, FFR was significantly higher in the group receiving standard dose versus high dose adenosine (0.89 [0.83–0.93 vs 0.87 [0.81–0.91], p = <0.001). In the control group, this difference was reversed to significantly lower FFR in standard dose compared to high dose (0.82 [0.75–0.85] vs 0.83 [0.77–0.89], p = 0.02) (Fig. 7). In three patients (4%), the higher dose of adenosine caused a change in agreement due to lowering FFR below the treatment threshold of 0.80 (0.85–0.79, 0.81–0.78, 0.81–0.79. The high dose did not decrease FFR below 0.75 in any of the 75 cases.

Following the landmark trials FAME [4] and FAME2 [3] the perception of assessment in non-critical coronary stenosis have been altered. FFR became a powerful tool not only for clinical practice, but also as the baseline standard in clinical research [42, 43]. The critical prerequisite of hyperemia and the side effects following systemic response of intravenous administration, have urged development of different agents, routes and doses. Even though different intravenous adenosine doses during FFR been investigated in an earlier smaller study [18], this present study concludes that their results indeed are reproducible in a larger, clinical patient population. Our results demonstrate that an increased dose adenosine do not improve accuracy in FFR, but is associated with increased side effects. Different pharmaceutical agents such as contrast media constitute an interesting option without the side effects of vasodilatation. The hyperemic capacity, mediated by osmosis, of contrast is well known, but has in previous trials been inferior to adenosine and papaverin [18]. Recent trials have demonstrated non-inferiority between contrast media and adenosine in the assessment of intermediate stenosis [17, 44], but even though this may be an option in certain lesions, there will still be need of adenosine-induced hyperemia in some cases; hence the question of increased adenosine doses needs to be adressed.

Adenosine interaction with A_2A receptor mediates a smooth muscle relaxation and thus a vasodilatation [45]. In addition, adenosine regulates the autonomic innervation in the heart by inhibition of the A₁ receptors, and by that, accomplishes a negative chronotropic and dromotropic effect in the conduction system [46, 47]. A concern in administration of adenosine, foremost in high dose regime, is that patients will develop adenosine induced systemic hypotension and bradyarrhytmias. In a recent trial, the predictor of hypotension during hyperemia was obesity and decreased microcirculatory resistance [48]. Of the original enrollment of 87 patients, a total of 11.3% developed an atrioventricular block during standard dose administration, and another 5.3% during high dose. Even though bradycardia usually is not persistent, it may create discomfort, which is an important reason why adenosine is prematurely terminated during FFR and hence, the hyperemic effect remains uncertain. Hypotension was uncommon in our study. Hypothetically, patients with unstable angina/NSTEMI and more complex lesions might be more prone to develop hypotension. Also, we cannot rule out that the intermittent monitoring of mean arterial pressure could affect the hemodynamic measurements by terminating infusions before severe hypotension develops. However, the vasodilation capacity of adenosine, displayed by shortness of breath, flushing and severe discomfort, was prominent. This observation confirms that symptoms during adenosine administration are not solely related to a systemic response or coronary hemodynamic changes, but rather via a direct effect on C-fibers in the lungs [49].

In this study, caffeine consumption was not randomized or controlled, but in a secondary analysis we found trends coherent with recent findings [31]. Matsumoto et al. [30] investigated the effect of intravenous adenosine doses of 140, 175 and 210 μg/kg/min in relation to papaverine during FFR. The effect of caffeine was less prominent at 220 μg/kg/min. The increased FFR results in patients on caffeine compared to controls in our trial may indicate an attenuated adenosine effect by coffee. However, the effect of increasing the dose of adenosine was marginal.

The present study has some limitations. The higher dose was always administered after the lower dose and a cross-over design might have yielded different results. However, this is the way adenosine is used in clinical practice, and no one would consider starting with the higher dose in all patients. The dose and timing of caffeine consumption were neither randomized nor controlled which might generate confounders. This was, however, an exploratory secondary endpoint. In terms of adenosine administration, we used a peripheral intravenous line, which compared to a central vein might have a slightly delayed systemic effect. However, we did flush the adenosine infusion together with an infusion of saline, a validated method to increase bioavailability and making the possible delay negligible. Also, in the era of transradial approach, a peripheral route for adenosin is desirable, and has been demostrating similar results in recent findings [50].

A considerable number of stable patients with type A lesions were enrolled. Although this reflects standard use of FFR in a real-life patient cohort, the possibility that a population with more advanced disease would have different results, cannot be ruled out.