In the heart, the purine nucleoside adenosine acts on heart rate, contractile function and coronary blood flow. Adenosine acts as a negative chronotrope in the sinoatrial and atrioventricular nodes and in atrial myocytes through adenosine A1 receptor activation [43, 45]. Adenosine acts as a negative inotrope, largely through its anti-adrenergic action after adenosine A1 receptor activation [43]. Adenosine acts as a coronary dilator, largely through activation of adenosine A2A receptors [43, 45]. In humans, arteriolar dilation in response to adenosine is mediated by A2A receptors, adenylate cyclase and calcium-activated potassium channels in vascular smooth muscle cells [52]. Also, indirect effects contribute to the role of adenosine in coronary blood flow: adenosine inhibits presynaptic release of norepinephrine from sympathetic nerves [43], inhibits platelet aggregation [35] and inhibits leukocyte adherence to the vascular wall [59]. A1 receptors are coupled to Gi/o proteins and their activation decreases cAMP, whereas A2A receptors are coupled to Gs proteins and mediate an increase in cAMP [1, 43].

Endogenous adenosine is an autacoid which is released with increases in heart rate and ventricular function during sympathetic activation. Originally, adenosine was proposed to be the substance which, through its coronary dilator action, served to mediate metabolic coronary regulation, i.e. the matching of coronary blood flow to the energetic requirements of the contracting myocardium [6, 17]. This original idea of adenosine as the mediator of metabolic coronary blood flow regulation was attractive, because adenosine is a catabolite of ATP, and thus could provide a stoichiometric coupling of energy demand and supply. Today, we have come to realize that cardiac adenosine metabolism is much more complex. In addition to the production of adenosine in cardiomyocytes, significant fractions of adenosine are continuously formed in the endothelium and at extracellular sites [12]. Adenosine is only one out of a number of redundant mechanisms which provide metabolic regulation of coronary blood flow, together with nitric oxide, activation of ATP-dependent K-channels and others [3, 13, 15]. During myocardial ischemia and reperfusion, adenosine is an important trigger and mediator of cardioprotection through activation of adenosine A1, A2B and A3 receptors [1, 11, 28]. Adenosine is also an algesic substance and involved in pain sensation during myocardial ischemia [58]; finally, adenosine attenuates left ventricular hypertrophy [55]. Thus, adenosine serves as a general feed-back mediator in the heart, certainly during pathological circumstances such as hypertrophy, hypoxia and ischemia and possibly also under physiological circumstances, e.g. by inhibition of the interaction of leukocytes with the endothelium.

Apart from its use to suppress supraventricular tachycardia, exogenous adenosine is used in patients with coronary heart disease mainly for two puposes: (a) to determine coronary reserve, and (b) to induce a maldistribution of myocardial blood flow into areas with persistent dilator reserve at the expense of areas with pre-existing compensatory vasodilation and thus reduced dilator reserve, i.e. a collateral and/or transmural steal [56, 61] and to image such maldistribution [7]. For the latter purpose, i.e. pharmacological stress testing, it is not important whether adenosine causes truly maximal coronary dilation as long as it acts as a powerful dilator. For the determination of coronary reserve, however, the assessment of true maximal coronary dilation is essential, and here is where myth and misconceptions exist.

The angiographic assessment of luminal diameter provides only incomplete information on the functional severity of a coronary stenosis, particularly with intermediate stenoses. To better estimate the importance of a coronary stenosis for myocardial perfusion, clinicians have taken resort to what they call “coronary physiology” [34]. The coronary pressure–flow relationship, which is determined in experiments in dogs and pigs at stepwise variations of pressure and/or flow, best characterizes coronary hemodynamics [8, 20, 23, 42, 54, 57] (Fig. 1): There is a zero flow–pressure intercept, i.e. coronary blood flow ceases at a perfusion pressure of about 20 mmHg; in the beating heart, both capacitance and inertial effects with storage of blood volume in epicardial vessels during systole and its discharge into subendocardial vessels during diastole contribute to such zero flow–pressure intercept. Under physiological circumstances, coronary blood flow then almost linearly increases when pressure increases from the zero flow–pressure intercept to about 40–50 mmHg. With further increasing pressure, autoregulation dampens any further increase in coronary blood flow such that in a range of pressures from 70 to 130 mmHg coronary blood flow is largely pressure-independent. The level of this autoregulatory plateau is set by myocardial performance and metabolism, i.e. coronary blood flow is higher with higher metabolism and vice versa [42] (dashed lines in Fig. 1). With a hypothetical maximal dilation, i.e. elimination of any active vasomotor tone, autoregulation is abolished and coronary blood flow is, at any pressure above the zero flow–pressure intercept, linearly related to perfusion pressure; in fact, since the coronary vasculature is not a rigid tube but distensible, the relation is even slightly concave away from the pressure axis. The pressure–flow relationship does not characterize the instantaneous relation of flow and pressure within a cardiac cycle but mean values of both flow and pressure during a hemodynamic steady state.

Fig. 1
figure 1

Coronary pressure–flow relationship. The curves are not hypothetical, but synthetized from original data in Refs. [2, 8, 20, 42, 54, 57], all in dog or pig hearts

It is clearly not feasible to generate a pressure–flow relationship in patients with coronary heart disease. The concept of coronary reserve condenses a major part of the information provided by the pressure–flow relationship into a single numerical value; coronary reserve is the dimensionless ratio of maximal coronary blood flow over baseline coronary blood flow [10, 18, 19]. A closer look at the pressure–flow relationship (Fig. 1) immediately identifies two problems of the coronary reserve concept: (a) the ratio of maximal over baseline coronary blood flow depends on perfusion pressure, and (b) baseline coronary blood flow and therefore also the ratio depends on myocardial performance and metabolism—notably on heart rate [41] as the single most important determinant of myocardial metabolism[26]. Attempts to correct for pressure and heart rate have therefore been made, by calculating resistances according to Ohm’s law (mean pressure/mean flow) [36], by relating coronary reserve to the pressure–rate product [62], or by normalizing coronary reserve in the vessel under study to a purportedly normal reference vessel in the same individual (relative coronary reserve) [4]. These attempts for correction, in turn, have problems: pressure and flow are not linearly related throughout and do not pass through the origin; therefore Ohm’s law is not really applicable. The error introduced by the use of Ohm’s law is the greater, the lower perfusion pressure and the closer it is to the zero flow–pressure intercept, i.e. in the presence of coronary stenoses. The pressure–rate product is only an incomplete estimate of myocardial metabolism and assumes homogeneous conditions throughout the entire heart. However, in coronary heart disease myocardial function, metabolism and perfusion are heterogeneous, and regions with normal flow and function co-exist with regions which have matched or mismatched reductions in flow and function [24, 32]. Substitution of the pressure–rate product would therefore not provide a valid estimate for regional myocardial metabolism, particularly not in poststenotic myocardium. In a patient undergoing invasive diagnostics for coronary heart disease the existence of an entirely normal vessel is unlikely, since even the presence of a risk factor is already associated with impaired coronary vasodilation [66, 67].

Notwithstanding the above limitations, the concept of coronary reserve relies on maximal coronary vasodilation, i.e. the elimination of all active coronary vasomotor tone, and then assumes that flow at any given pressure is solely determined by the morphology of the coronary vascular bed and by extravascular compression [3, 49], which is again largely determined by diastolic duration and thus heart rate [33, 51].

Is adenosine or any other dilator substance indeed capable of eliminating all active vasomotor tone? Increases in coronary blood flow from about 1 ml min−1 g−1 at baseline up to 5 ml min−1 g−1, i.e. a coronary reserve up to 5, with individual values exceeding 8, have been reported using intracoronary Doppler flow measurements in patients with normal coronary angiography undergoing invasive diagnostics for chest pain [65] and in normal volunteers using positron emission tomography [9]. When using Doppler measurements of coronary blood flow velocity only, coronary reserve is probably somewhat underestimated, since a small additional epicardial coronary dilation is neglected. Intracoronary Doppler data are not available from true normals, since there has always been a history of chest pain, and even the presence of a risk factor is associated with impaired vasodilation [66, 67]. Also, a thorough exclusion of atherosclerosis by intravascular ultrasound in the patients with normal coronary angiograms [16] is missing in most studies. Therefore, the highest reported values for maximal coronary blood flow and coronary reserve come most likely closest to true physiological values, since all confounding factors tend to induce an underestimation of maximal coronary blood flow. To assure that indeed a maximum increment in coronary blood flow is recorded, incremental doses of adenosine must be given, and ultimately a lack of further flow increase with an additional dose of adenosine documented [65]; this is only rarely done in clinical diagnostics [14, 44], and mostly standard adenosine doses are given, normalized for coronary artery perfusion territory on intracoronary use or for body weight on intravenous use [34]. Adenosine has a fair selectivity for the coronary circulation, but at systemic doses aiming for maximal coronary vasodilation, hypotension and reflex tachycardia are inevitable, both of which impair the coronary blood flow response. Also, the algesic action of adenosine may limit its use at higher doses in some patients. On the other hand, the use of adenosine is safe, since it is rapidly eliminated from the vascular space by uptake into erythrocytes and endothelium [12].

Even high-dose adenosine does not eliminate all coronary vasomotor tone. α-Adrenergic coronary vasoconstriction is not eliminated in the experiment [30], and coronary reserve in humans is limited by both α1-adrenergic and α2-adrenergic coronary vasoconstriction [5, 21, 27, 39, 50]. The limitation of coronary reserve by α-adrenergic coronary vasoconstriction is particularly important with coronary atherosclerosis [5] and during percutaneous coronary interventions [21, 22]. In the experiment, adenosine is also not capable to eliminate endothelin-induced coronary vasoconstriction [38], and it is unclear to which other coronary vasoconstrictors (angiotensin, serotonin [37], thromboxane A2 [37], vasopressin) the inability of adenosine to fully eliminate vasoconstriction pertains.

Are there alternatives to adenosine to induce maximal coronary dilation? Reactive hyperemia responses following 20-s coronary occlusion resulted in a coronary velocity reserve value of 5.8 in patients undergoing cardiac surgery but with normal coronary angiograms [40]; this mean value appears to be somewhat higher than those reported for adenosine. Dipyridamole, which inhibits the nucleoside transporter in cell membranes and then potentiates the action of endogenous adenosine [53], induces a coronary dilation of similar magnitude as that by adenosine in the “normal” coronary circulation [9, 63, 64]. The same is true for intracoronary papaverine [41, 64, 65], but papaverine induces undesired EKG abnormalities [34]. In a thorough study with definition of normal coronary status not only by angiography but also by intravascular ultrasound, intracoronary papaverine recruited a coronary reserve of 5.3 [16]. Intracoronary nitroprusside caused equivalent coronary vasodilation to adenosine, yet at pre-existing subnormal coronary reserve [46]. In the experiment, dihydropyridine calcium antagonists can eliminate α-adrenergic coronary vasoconstriction [29] and still recruit a dilator reserve in ischemic myocardium [31], but their negative inotropic action might limit their use in the clinical assessment of coronary reserve.

Notwithstanding all of the above critical considerations and limitations, coronary reserve assessment using standard doses of adenosine and with cut-off values substantially below normal values, often below 3, provides important prognostic information and reasonably guides clinical decisions on coronary interventions [25, 34]. The concept of fractional flow reserve entirely relies on the use of Ohm’s law and is solely based on coronary pressure rather than flow measurements to assess coronary dilator capacity [48]; despite thus being even further away from the physiology of the coronary pressure–flow relationship, fractional flow reserve assessment is more easily feasible and empirically validated to provide prognostic information [47, 60].

In summary, to identify and contradict the most frequent misconceptions: (1) Adenosine is not the predominant mediator of physiological regulation of coronary blood flow. (2) Adenosine does not eliminate all active coronary vasomotor tone. (3) Coronary reserve is not independent of hemodynamics, notably arterial pressure and heart rate. (4) A coronary reserve of 3 is not normal. Thus, the physiological assessment of coronary reserve with adenosine is really a myth: it has a true core which is overshadowed by layers of pragmatic compromises, and just because this myth is so often told, truth and fiction are often no longer distinguished in daily practice. In conclusion, I am not opposing empirically founded standard use of adenosine and coronary reserve cut-off values as low as 2.7 for pragmatic clinical decision-making, but I want to raise the awareness that such procedure is far away from coronary physiology.