This is the first study evaluating the influence of caffeine on the ischemic burden by a two-exam adenosine stress CMR protocol including both a caffeine-naïve CMR scan and a repeat CMR scan after defined caffeine intake. Major findings are: 1) Despite intake of caffeine prior the repeat CMR exam, no conversion of an ischemic-positive to an ischemic-negative study could be observed in this high-risk population. 2) Although significant differences in the extent of ischemic burden were demonstrated between caffeine-naïve and post-caffeine CMR exams based on a 16-segment model, 60-segment model and total ischemic volume, differences were small in absolute terms, and no prognostic relevant myocardial ischemia (≥2 segments in a 16-segment model) was missed despite consumption of caffeine [
26]. 3) A history of CAD seems to have no influence, since no significant differences could be observed for patients with prior myocardial infarction, known CAD or previous CABG vs. patients with no history of CAD. 4) No correlation could be found between serum caffeine levels and the number of ischemic segments.
Myocardial perfusion defect by segment and total volume
Our observation that ischemic burden tends to be reduced after prior caffeine intake could be confirmed by quantitative analysis of the 16-segment model: 7.9 ± 3.5 segments showed myocardial ischemia without caffeine vs. 6.9 ± 3.5 segments after caffeine intake,
p < 0.001. Therefore, despite not missing any relevant myocardial ischemia under caffeine, there seems to be attenuation of ischemic burden induced by the presence of caffeine. This effect might be in part explained by the caffeine dose of 200 mg, which is known to represent “significant” amount of caffeine [
19]. In contrast, Zoghbi et al. [
14] studied the effect of an 8-oz. cup of brewed caffeinated coffee (with a caffeine content varying from 25 mg to 240 mg) one hour before adenosine gated SPECT. Consequently, the latter study [
14] demonstrated lower caffeine levels ranging from 3.1 ± 1.6 mg/L, whereas in the present study patients showed caffeine levels in the range of 4.6 ± 2.2 mg/L, suggesting a distinct effect of caffeine on coronary hyperemia. Our results are in line with a study from Namdar et al. [
9]. They studied the effect of 200 mg caffeine (equivalent to our dose) on myocardial blood flow at rest and exercise in healthy volunteers at normoxia and during acute exposure to stimulated altitude by
15O–labeled H
2O and positron emission tomography. They found that a dose of two cups of coffee (200 mg caffeine) significantly decreased exercise-induced myocardial blood flow at normoxia and at hypoxia, suggesting that exercise-induced hyperemic flow response may at least in part be antagonized by caffeine [
9].
Most of our patients had ischemic burden comprising several myocardial segments, Figs.
2,
3 and
4. Data analysis revealed 7.9 ± 3.5 ischemic segments without caffeine vs. 6.9 ± 3.5 ischemic segments with caffeine, identifying our patients cohort as a subset of very-high-risk patients, since another group could show [
29] that patients with >5 ischemic (of 16) segments had a risk of an adverse CAD event of approximately 14%/year. Although our results emphasize that on a 16-segment model basis the influence of caffeine leads to a decrease of ischemic burden by only 1 segment (7.9 ± 3.5 vs. 6.9 ± 3.5), these differences were significant (
p < 0.001). One might argue that for patients further management, there won’t be a big difference if eight or seven myocardial segments were involved, postulated that these ischemic segments are supplied by the same coronary artery. However, studies suggest that the unadjusted hazard for CAD, death or myocardial infarction was elevated approximately by 1.2 for every segment with an ischemic perfusion defect compared to patients with a normal stress study who have an observed CAD event rate of approximately 1%/year [
26,
30,
31]. Therefore, caffeine-induced effects on myocardial ischemia might mask not only patients’ accurate diagnosis but also his prognosis.
Interestingly, one of our patients had only 2 ischemic segments in his caffeine-naïve exam, which is considered as a threshold for moderate-severe myocardial ischemia [
26], indicating adverse outcome which might warrant further invasive diagnosis by coronary angiography. In this case, despite intake of caffeine in the follow-up exam, the repeat scan still demonstrated 2 ischemic segments, pointing towards the hypothesis that no prognostic relevant myocardial ischemia is missed by prior caffeine intake in an adenosine stress CMR test. However, in cases in which only 2 of 16 myocardial segments are involved, prior caffeine might substantially increase the risk of a false-negative stress CMR or at least the probability to detect myocardial ischemia in just a single instead of two segments. At first sight, this might be a negligible difference. However, it is of clinical importance to detect the true extent of ischemic burden not only for diagnostic but also for prognostic purposes [
26,
32], since patients with zero or just one ischemic segment can be safely deferred from revascularizations and show a favorable outcome on medical treatment that does not differ from those patients with normal CMR perfusion studies [
32]. Based on our findings, we should expect, that in lower ischemic burden patients significant ischemia would be missed, affecting not only prognostic assessment but the diagnosis itself.
In the 60-segment model, we found an even higher difference in ischemic segments between caffeine-naive and caffeine adenosine CMR stress scans: 18.6 ± 8.7 vs. 15.7 ± 8.7 segments, p < 0.001. This illustrates that even in a very detailed model results are comparable to the results of the 16-segment model by showing slight, but significant differences in the absolute number of ischemic segments induced by prior caffeine consumption.
Likewise, total ischemic volume between caffeine-naïve and caffeine-consumed stress scans demonstrated slight but significant differences: 4.2 ± 2.5 ml vs. 3.4 ± 2.4 ml, p < 0.001, Figs.
2,
3 and
4, underlining a low but significant impact of caffeine on the extent of ischemic burden in adenosine stress CMR tests.
Patient subgroups
Subgroup analysis revealed that patients with no LGE compared to patients with ischemic LGE demonstrated no major differences in the number of ischemic segments in their initial vs. their repeat CMR scan with regard to the 16-segment model,
p = 0.89, the 60-segment model,
p = 0.46, as well as for the total quantified ischemic volume,
p = 0.37, Table
3. This is of importance since one might argue that LGE in patients might interfere with the potential extent of myocardial ischemia especially in terms of fixed perfusion defects. Similar results could be observed for the ischemic burden between the initial and repeat CMR scan in patients without a history of CAD vs. patients with known CAD with regard to the 16-segment model (
p = 0.43), 60-segment model (
p = 0.73), and total ischemic volume (
p = 0.85). Likewise, in patients with no previous coronary artery bypass graft (CABG) vs. patients with prior CABG, segments of ischemia in a 16-segment model, 60-segment model, and total ischemic volume demonstrated no significant differences between both CMR exams (
p = 0.58,
p = 0.18,
p = 0.26, respectively). These results underline that the presence of caffeine itself seems to be the main driver of a reduced ischemic burden, independent from patient’s cardiac history.
Caffeine levels and ischemic segments
Despite varying caffeine levels after similar caffeine intake (200 mg each) at our institution, no correlation could be found between caffeine levels and the number of involved ischemic segments. This is in line with Lee et al. [
13], stating that the concentration of caffeine (at baseline or after supplementation) was not associated with percent defect reversibility. Furthermore, the amount of change of caffeine levels from the initial CMR to the second CMR exam after caffeine consumption had no effect on percent defect reversibility,
p = 0.97. Reyes et al. [
33] investigated 30 patients with known or suspected CAD with and without caffeine by clinically indicated myocardial perfusion imaging. They found that myocardial ischemia decreased by presence of caffeine with the standard use of 140 μg adenosine but did not change significantly with the use of the higher adenosine dose of 210 μg suggesting that in patients with prior caffeine consumption the protocol might be switched to the higher adenosine dose. The reason for this finding might be the competitive interaction between adenosine and caffeine, so receptor blockade by caffeine could be surmounted by an increased dose of adenosine. However, the higher dose is not approved for use in the United States in imaging [
1].
Limitations
Since this is a single-center study, potential center-specific bias cannot be excluded. Furthermore, the results of this study were raised in a population with extensive ischemic burden, and might not be transferred to patients which demonstrate an ischemic burden comprising only 1 or 2 myocardial segments. Therefore, our results cannot be generalized to all patients with CAD. Furthermore, quantification of ischemic burden by a 2D 3-slice approach may be inferior to a 3D full coverage approach. However, our 3-slice approach is common practice for clinical routine, and underlines the real-world character of this study.
We have not addressed the ingestion of different caffeine amounts in order to detect a potential threshold at which caffeine shows definite impact on the extent of ischemic burden. However, intention of our study was to reach significant serum levels of caffeine to demonstrate the influence of caffeine on myocardial ischemic burden. Furthermore, a previous study from Lee et al. [
13] assessed adenosine-induced myocardial perfusion imaging defects over a broad range of caffeine concentrations with SPECT, and found no significant caffeine effect.
Moreover, most of the aforementioned studies are performed with SPECT since there is only scarce data about the effect of caffeine on adenosine stress CMR, which is known to have better spatial resolution than SPECT. Therefore, not all the data might be applied to the technique of CMR.
In this study, coronary angiography was used as the gold standard for the detection of significant CAD. Nevertheless, one should keep in mind that the sole anatomical presence of a stenosis does not always provide sufficient information regarding its hemodynamic relevance. Thus, functional assessment by intracoronary pressure wire (FFR) or intravascular ultrasound studies would have been highly desirable, but was not carried out in this study.