The concept of coronary flow reserve (CFR) was introduced to reflect the maximal oxygen delivery capacity during increasing demand. Since cardiomyocytes are capable of extracting oxygen with great efficiency already at rest, the only way of meeting increasing demand is to increase coronary blood flow [11, 12]. At rest, there is a significant autoregulation in the coronary vascular bed which ensures sufficient blood supply to the cardiomyocytes under a wide range of perfusion pressure. Thus, resting coronary blood flow is mainly under control of tissue metabolic demand [13]. To test the maximum flow reserve capacity, the autoregulation must be uncoupled, which will create a linear relationship between perfusion pressure and flow. Thus, CFR is usually measured during either exercise challenge or pharmacological stimuli such as dobutamine and adenosine [14].

Hemodynamically significant coronary lesions are known to cause reduced CFR [12, 14]. In the case of diffuse coronary artery atherosclerosis without angiogram-visible significant stenosis, CFR could be an objective tool to assess the hemodynamic significance, not necessarily reflected by distinct local stenosis. This has been comprehensively described by Dr. Gould et al., who pioneered the concept in humans [15]. However, impaired CFR could be due to many other pathological conditions, such as microvascular dysfunction in for example: diabetic patients [16]. Interestingly, the presence of thin cap fibroma in the coronary artery has been reported to be associated with impaired CFR, indicating an important but not completely surprising link between atherosclerotic lesion composition and downstream vascular function [17].

While FFR seems to be a fit-for-purpose tool to address hemodynamic consequences of focal lesions, CFR could be a measure of the entire coronary vascular functionality. Even though there is correlation between CFR and FFR, significant discrepancy between these two measures is evident, which may suggest that they indeed reflect distinct aspects of the coronary arteries [18]. This means that not all patients with impaired CFR would have any intervenable coronary lesions according to FFR, and that CFR will not necessarily be improved even following the best FFR-guided PCI procedure.

Even though myocardial perfusion scintigram (MPS) is considered the gold-standard for assessment of myocardial ischemia, the method relies on flow-redistribution during stress testing when a hemodynamically significant stenosis is present. In the case of balanced three vessel disease and microvascular dysfunction, MPS results are less reliable in term of identifying the high risk patient [19]. CFR as an objective quantitative tool, will assess the hemodynamic significance of angiographic moderate or ambiguous stenosis, as well as diagnose microvascular diseases (Fig. 1) [15].

Coronary flow reserve can be measured using many modalities. Thermodilution was one of the invasive techniques used to determine CFR in man [20]. Coronary flow velocity reserve (CFVR) can be measured using invasive Doppler-guide wires using, for example adenosine as a stress agent [21]. Recently, a combined FFR and CFVR catheter has been launched for simultaneous flow and pressure measurement [22].

Non-invasive techniques have been developed during the last two decades taking advantage of modern high-end imaging modalities. Positron emission tomography (PET) is now established as the gold standard for absolute coronary blood flow and flow reserve measurement. PET provides both ischemia evaluation as well as absolute flow quantification [23]. Various tracers can be used including those that do not require a full-scale cyclotron facility, for example: rubidium [24]. In combination with high-resolution computer tomography (CT), this modality can provide both structural and functional information about ischemic heart disease [25]. Even though PET is an established approach to assess CFR, radiation exposure as well as high equipment and running cost limit its wide use. Contrast CT might be a new modality for perfusion imaging. This method is also associated with radiation although the level of radiation has decreased substantially with the latest generation CT [26]. Magnetic resonance imaging is a non-radioactive valuable method, capable of assessing CFR, but is relatively costly and time-consuming [27].

Myocardial contrast echocardiography is an inexpensive and non-radioactive method for measurement of CFR. When image quality is adequate, CFVR as well as CFR can be readily measured [28]. However, for accurate and reproducible quantification of CFVR and CFR, high-quality images are required both during rest and hyperemia allowing contrast intensity measurement to derive flow velocity as well as steady-state flow signals.

Transthoracic color Doppler echocardiography (TDE-CFR) has been successfully used to measure coronary blood flow velocity (CBFV) and CFVR [29]. With proper machine setting and skilled hands, the technique is non-radioactive and feasible in every echo-lab without need of contrast agent. Using this approach, success rate is greatest in the left anterior descending (LAD) coronary artery. In addition, right posterior descending coronary artery [30] as well as circumflex artery (LCX) CFBV is measurable [31]. Accuracy of these approaches has been validated previously against invasive techniques [30]. Technically, for access to the distal LAD, only a small echo-window is needed compared to for example: left ventricle ejection fraction measurement using Simpson’s rule, where a couple of cm echo-window is needed for good image quality. Thus, obesity is not an absolute hurdle for the success of this technique [32]. For real-life clinic use of this technique for diagnosis of ischemic heart disease including obstructive CAD and microvascular disease, dipyridamole-assisted TDE-CFR has been successfully applied on a routine basis in combination with wall motion analysis [33]. Using this approach, the high specificity of stress echo can be combined with the high sensitivity of CFR for diagnosis of ischemic heart disease in patients with suspect CAD.

For experimental purposes, LAD TDE-CFR has been used for risk prediction as well as documentation of intervention effects. Recently, our group showed excellent inter- and intra-day variability of CBFV in a randomized double blind clinical study [34]. During this study CBFV upon increasing doses of adenosine was measured repeatedly. To minimize environmental influences, all subjects stayed at the clinical trial unit the evening prior to the examination and received standardized meals etc. Thus, under controlled conditions, variability for CFBV measurement using this approach was satisfactorily low (Coefficient of variance for AUC for CBFV and adenosine doses <1 %) (Fig. 2).

To further understand biological processes and the clinical relevance of CFR, numerous studies have been performed to relate CFR to cardiovascular biomarkers. These studies revealed that CFR indeed could serve as a composite physiological marker of systemic CV risk. Besides obstructive CAD, which is an obvious determinator of CFR [12, 14], impaired CFR has been reported in patients with dilated cardiomyopathy [35], as well as patients with diabetic vascular complications [36]. Further, myocardial capillary density has also been shown to determine CFR in patients with dilated cardiomyopathy [37].

CFR is rheologically associated with blood viscosity, which in turn is dependent on lipoprotein and fibrinogen levels. In familiar hypercholestoremic patients, CFR is related to LDL level and is improved after LDL apheresis, supporting the concept that increased blood viscosity due to extremely high LDL level can be a determinant of CFR [38].

CFR has also been shown to correlate to markers of low degree systemic inflammation, such as C-reactive protein [39] and neutrophil count [40]. Based on large-scale genetic association studies, IL-6 has recently been shown to be causally linked to CV outcome alongside LDL-regulating pathways [41]. Interestingly, a PET-CFR study conducted in twins showed that IL-6 indeed was a determinant of CFR in those asymptomatic subjects [42].

It is well-known that endothelium plays a central role in CV disease etiology and progression [43]. Even though CFR is measured using a non-endothelium dependent vasodilator, for example: adenosine, for achieving maximum hyperemic flow response, intact endothelial function is required. CFR has been shown to be associated with peripheral endothelial function both in chest pain patients and in healthy volunteers [44, 45] and direct inhibition of the nitric oxide pathway reduced hyperemic response in healthy volunteers [46].

Impaired CFR has been shown to confer great prognostic information for hard CV outcome, including all cause mortality, CV death and non-fatal MI. Independent of imaging modality used, the prognostic power of CFR appears to be consistent. In post-MI setting, invasively assessed CFVR showed prognostic value in the “DEBATE” study. In such a setting, microvascular obstruction as well as infarct size may impact on CFVR alongside residual ischemia [47]. This concept has been extensively evaluated in patients with suspected coronary artery disease since they are undergoing ischemia testing based on clinical indication. In these patients, CFVR adds incremental prognostic values on top of wall motion analysis in terms of prediction of mortality and spontaneous CV events [33]. Interestingly, in the paper by Murthy et al., PET-CFR magnitude dependently predicted CV mortality across various levels of infarct size, ischemic burden and LV ejection fraction [48]. These data strongly supports the concept that CFR rather than just obstructive coronary artery disease contributes to CV vulnerability. In line with this, Murthy et al. showed that diabetic subjects without previous CAD but with impaired CFR had the same risk as non-diabetic patients with CAD [49]. Further, CFR showed great prognostic information in patients with dilated cardiomyopathy [50], which suggests that impaired CFR is an independent risk driving CV events in additional patient cohorts. In the context of post-infarct cardiac remodeling, impaired CFR is an independent predictor of adverse left ventricle remodeling [51]. The quantitative nature of CFR measurement is also reflected in its prognostic information, since it has also been shown that CV outcome is related to CFR in a magnitude dependent manner, that is the lower the CFR the worse the outcome and vice versa [48].

Transplant vasculopathy is a phenomenon of diffuse coronary artery disease for which traditional angiogram-based approaches cannot be used to follow the disease progression. In these patients, CFVR has been successfully used as a prognostic marker [52]. This further supports the concept that the integrative physiological assessment of the coronary functionality is of greater importance than only the luminology-based approaches.

That surgical by-pass or removal of hemodynamically significant obstructive epicardial lesions can improve CFR is of course a direct consequence of these procedures and follows the purpose and definition of a successful revascularization [53]. Other established CV treatments have also been shown to improve CFR through mode of action impacting on mechanisms controlling CFR (Table 1).

Discovery and development of novel anti-atherosclerotic drugs to reduce cardiovascular events on top of standard of care is an area of high risk. Several difficulties have to be overcome in order to succeed with delivering a positive outcome in phase III trials. Traditionally, novel targets and compounds are validated in pre-clinical models of atherosclerosis with the primary read-out being a reduction in plaque progression/size. Often such data provide the sole basis in terms of in vivo intervention data before investing in large clinical programs. For clinical proof-of-concept, primary end-points have focused on plaque centric endpoints such as plaque volume, intima-media-thickness or plaque composition. Although these endpoints are associated with CV disease and CV events, limited information is available on how intervening with these endpoints affects CV benefit. This is not surprising since, atherosclerotic plaques are only one component contributing to the ischemic myocardial event in humans. Many additional biological mechanisms have been shown to play a role in triggering the event including microvascular dysfunction, endothelial dysfunction, systemic inflammation, vasospasm, thrombosis and angiogenesis [67‐70]. By using a more integrative approach for proof-of-concept both pre-clinically and clinically, clinical development could be potentially de-risked.

To study the effects of novel anti-atherosclerotic compounds in animals within reasonable time frames, disease progression needs to be accelerated by genetic manipulation and/or feeding high-fat high-cholesterol containing diet. No single animal model available today mimics all aspects of this complex human disease and it is important to note in particular that no model reliably and reproducibly replicates the rupture and subsequent thrombotic occlusion of coronary arteries. As discussed previously, the pathophysiological events that occur at the level of the atherosclerotic plaques have been of great interest during the past few decades and also in animal studies targets which are involved in reducing plaque progression have been in focus for potential future interventions for reducing cardiovascular events. To gain confidence in potential novel therapies for reducing CV events, it is crucial that pre-clinical studies focus on the proper vascular bed and the relevant translational endpoints. Despite the prevalent use of mouse models of atherosclerosis, few have focused on the coronary arteries due to technical difficulties with studying these small arteries. Fortunately, due to recent advances with high-resolution imaging, it is now possible to perform morphological and functional imaging studies of cardiovascular systems in living mice [71‐74]. This has dramatically increased the translational value of preclinical disease models and may provide guidance for dose selection and Proof of Principle studies in humans.

Despite the high heart rate in mouse, coronary flow velocity can be measured in a reproducible way using high-resolution color Doppler ultrasound. This technique has now been applied in rodents and rabbits according to the human protocol (Fig. 3). The flow profiles in mouse left coronary artery (LCA) are strikingly human-like regarding amplitude, duration and systolic/diastolic distribution [71]. Due to its small size, little is known about coronary atherosclerosis and its functional implications in mice. By using ultrasound biomicroscopy the LCA can be visualized and the extent of atherosclerosis determined in a non-invasive, longitudinal way in living mice [75]. In a study performed in apoE/LDL receptor-deficient mice the percent wall thickness in the proximal LCA increased with age. Percent wall thickness was significantly associated with color Doppler echocardiography assessed mid- to distal coronary flow velocity ratios (Fig. 4). The extent of coronary atherosclerosis in the proximal LCA was confirmed by histological sections and correlated well with percent wall thickness and color Doppler echocardiography velocity ratios.

Our group pioneered the work to establish the TDE-CFVR approach in mice [71, 72, 74, 75] using high-resolution ultrasound. In addition to mid- to distal flow velocity ratio approach to determine LCA stenosis, CFVR can be determined in mice in a similar way as in humans using color Doppler echocardiography. Hyperemia can be induced using adenosine at 160 μg/kg/ml which is similar to the dose used in humans. This dose was shown to induce hyperemia without any apparent effects on systemic hemodynamics. Hypoxia is an alternate non-invasive means of inducing a similar increase in CFVR. In LDLR-deficient mice with coronary atherosclerosis, a correlation could be found between the extent of lumen narrowing and CFVR (Fig. 5) [72]. Although a correlation between coronary atherosclerosis and CFVR was demonstrated in mice it has also been shown that inflammation could impact on CFVR, for example: CFVR is reduced following experimental coxsackie virus-induced myocarditis [73].

Although similar protocols are used to study CFVR in mice and humans, caution should be taken when comparing the absolute magnitude of changes since anesthesia is necessary in animal experiments and this has been shown to influence CFR in rodents [76]. Indeed, the most commonly used anesthetic agent isoflurane has been to shown to increase CFBV per se [77]. To assess coronary artery function in mice, CFVR is a good surrogate for CFR. In humans, a good correlation between CFVR and CFR has been demonstrated [78]. By measuring LCA lumen diameter and flow velocity simultaneously in mice, a good correlation between CFVR and CFR was evident. However, due to the larger intra- and inter-observer variability in CFR measures, CFVR is a more robust marker of coronary vascular health in mice [74] (Fig. 6).

The TDE-CFR approach has also been developed and applied in other species, for example: rat, to demonstrate effects of various interventions. We showed that spontaneous hypertensive rats improved their CFVR following voluntary endurance training through increased vascular anti-oxidant capacity [79]. Improvement of CFVR was also associated with improved vascular compliance and acetylcholine-mediated vasodilation in peripheral vascular beds [80] (Fig. 7).

Statins have become a standard therapy for CV patients and any novel therapy will need to demonstrate significant benefit on top of statins. Thus, back translation of statin effects to animal models is of fundamental importance for drug discovery and development. In addition to their cholesterol lowering effects, statins have been shown to retard the progression of atherosclerosis [81], improve endothelial function [82], reduce systemic inflammatory markers and reduce cardiovascular mortality [82‐84]. The response to statins in animal models varies with species, strain, gender, diet and protocol used. Although statins reduce plasma cholesterol and plaque area in some murine models, the exact mechanisms behind the cholesterol lowering effects are largely unknown. Statins have shown to reduce atherosclerosis in the apoE*3Leiden by changing lipoprotein profiles and total lipid levels [85, 86] in plasma. However, the so called “pleiotropic effects” may also play a significant role in this model [87]. In apoE- and LDLr-deficient mouse strains, statins have shown variable results on plasma cholesterol levels and plaque progression. ApoE-deficient mice on a mixed C57BLl/6 and 129SvJ (Taconic Denmark) background on a high-fat diet (21 % pork lard and 0.15 % cholesterol) responded to rosuvastatin in a human like manner. Plasma Cholesterol levels were reduced on average by 20 % during the 16-week study period [88]. In addition, brachiocephalic plaque area and the inflammatory serum biomarker serum amyloid-A (SAA) were significantly reduced at the end of the study suggesting additional effects similar to those observed in humans. Assessing coronary artery flow, calculated from UBM measured coronary diameter and color Doppler echocardiography flow velocity revealed an increase in hyperemic blood flow at the end of the study, again in line with findings in humans. Generating data in this animal model responding to statin therapy in a human-like way will improve our confidence in novel targets and facilitate future translational drug intervention studies in mice.

Grönros et al. used the TDE-CFVR approach to document effects of a novel arginase inhibitor, which may act by restoring normal nitric oxide balance in the vascular endothelium. In a rat model of type 2 diabetes, microvascular dysfunction as measured by CFVR was demonstrated relative to control rats. Further, the arginase inhibitor succeeded to improve CFVR only in diabetic rats and not in control rats [89].

We and others have shown that CFR in vivo is reduced in late stage mouse atherosclerosis mouse models [72, 90], but also in the absence of coronary arthrosclerosis such as in models of diabetes [91], blood pressure overload induced heart failure [92] and myocarditis [73]. CFR can thus be measured in various disease models in vivo for exploration of disease mechanisms and might be used to validate novel drug targets.

5-Lipoxygenase (5-LO), five lipoxygenase activating protein (FLAP) and its products leukotrienes B₄ (LTB₄) and cysteinyl leukotrienes (cys LTs) have been linked to cardiovascular disease [93]. The ALOX5AP (FLAP) haplotype HapA has been associated with risk of MI [94] and levels of 5-LO, FLAP and LTA₄H mRNA in human carotid plaques have been correlated with the occurrence of plaque instability [95]. In a phase II study, the 5-LO inhibitor, VIA-2291 reduced coronary plaque volume and new formation of rapidly growing coronary lesions in ACS patients following 6-months treatment [96]. These findings are in great agreement with many preclinical studies performed in animal models of atherosclerosis [97, 98].

To further explore the potential role of the 5-LO pathway in low degree subclinical inflammation and coronary vascular function, we carefully investigated CV function in 5-LOX-deficient mice (AstraZeneca Transgenic Centre, Mölndal, Sweden). We hypothesized that these mice could be protected from deleterious effects of lipopolysaccharide (LPS) challenge in a low-grade inflammatory, sub-acute setting.

Adenosine-induced CFVR, cardiac function and morphology were measured before and one week after LPS-challenge (2 mg/kg) in 5-LOX^−/− and in back-crossed littermate C57BL/6 wild-type mice. Several cytokines and ex vivo aortic endothelial function were measured one week after LPS challenge to avoid acute sepsis-like condition. Cardiac function (not shown) and CFVR did not differ between the groups before the LPS challenge. However, one week after LPS challenge, CFVR was decreased in WT mice (p < 0.05) while no significant change was evident in 5-LOX^−/− mice (Fig. 8). In vivo cardiac measurements were similar after LPS (not shown). Ex vivo aortic acetylcholine-mediated relaxation capacity in phenylephrine pre-contracted segments of the thoracic aorta was impaired in WT mice compared to 5-LOX^−/− mice post LPS (p < 0.01). Levels of IL-10 were higher in the plasma of 5-LO^−/− mice compared to WT mice post LPS (p < 0.05).

Higher doses of LPS might induce septic shock with acute effects such as myocardial contractile dysfunction, drop in systemic blood pressure, vasoplegia, tissue hypoperfusion, microvessel injury and multiple organ injury [99], whilst low or moderate doses of LPS have been shown to induce endothelial dysfunction in experimental settings [100]. Our data seems to be consistent with the previous findings that 5-LOX-deficient mice maintain blood pressure control better than control mice [101] and rats treated with a 5-LO inhibitor (Zileuton) showed reduced organ failure [99] following severe LPS endotoxemia. In addition to mediating inflammation [93, 102], direct vasoactive actions of the LTs have been frequently reported in both coronary [103] and aortic [101] arteries of several species. Thus, 5-LO metabolites seem to have both direct and indirect vasomodulating properties, especially in abnormal conditions, as observed in atherosclerotic arteries of humans [103]. Taken together, in addition to its anti-atherosclerotic effects, 5-LO pathway inhibition may also improve coronary vascular function in patients with systemic low-grade inflammation.

In summary, accumulating evidence indicates that CFR may serve as a composite marker for coronary macro- and microvascular status, as well as systemic risk profile. Quantitative measurement of CFR provides us with an accurate prognostic tool to risk stratify patients as well as to follow disease progression and potential effects of intervention. Mechanistically, CFR is closely associated with many major pathoetiological factors driving residual risk in CV patients, such as epicardial vessel atherosclerosis burden and composition, left ventricle hypertrophy, vascular rarefaction, dyslipidemia, subclinical inflammation, endothelial dysfunction, etc. Establishment of the translational imaging techniques opens up exciting possibilities for deepened mechanistic understanding of this important pathophysiological phenomenon. Accumulated knowledge will hopefully accelerate development of novel therapeutic approaches targeting unexplored pathways to improve CFR and thereby further reduce CV mortality and morbidity. A translational approach and comprehensive understanding of the biology may potentially de-risk clinical development programs of novel therapies. In addition to treatment correcting dyslipidemia and emerging drugs to reduce systemic inflammation, recent scientific progress within regenerative medicine may also in the not too distant future contribute to step-change therapies in terms of for example: regeneration of cardiomyocytes and coronary microvessels, and thereby restore a functional myocardium with well-dimensioned vascular bed.