*4.1. Biochemical Concepts*

In general, biochemical concepts are "prone to potentially harmful effects, since arteriogenesis shares many common mechanisms with inflammatory diseases, such as atherosclerosis" [44]. Accordingly, Epstein et al. coined (biochemical) collateral promotion a "Janus Phenomenon", that is, "whatever intervention enhances collaterals increases atherogenesis and vice versa" [57].

With the rapid development of angiogenic growth factors and the growing understanding of their mechanisms of action, multiple trials testing collateral growth promotion have been initiated. Because of the known pivotal role of monocytes in orchestrating the different processes of angio- and arteriogenesis [50], most of the projects focused on the activation or the recruitment of this cell line. Growth factors most extensively studied have been granulocyte-macrophage colony-stimulating factor (GM-CSF) [58–62], granulocyte colony-stimulating factor (G-CSF) [63–66] or monocyte chemoattractant protein-1 (MCP-1) [41]. Besides, also different fibroblast growth factors (FGF) [67–69] and VEGF [70] have been clinically tested. Altogether, this study showed that angiogenesis is less efficient than arteriogenesis [71] for promoting bulk collateral blood flow, since it only promotes microvascular density. Consequently, clinical trials evaluating the e ffect of angiogenetic factors such as FGF or VEGF have failed to demonstrate a therapeutic e ffect that exceeds the e ffect of placebo treatment [67,68,70].

Colony-stimulating factors, on the other hand, have been found to promote the formation of large interconnecting arterioles (arteriogenesis), which are required for the salvage of myocardium in the presence of occlusive CAD [58]. Buschmann et al. [61] found that a continuous infusion of GM-CSF into the stump of the acutely occluded femoral artery of rabbits enhanced blood flow to the hind limb five-fold. The mechanism of action in that study has been found to be the prolonged survival of monocytes, "known to play a decisive role in arteriogenesis" [61]. In two small but randomized and placebo-controlled clinical trials with 35 patients in total, GM-CSF has been shown to be e fficacious in a short-term subcutaneous administration protocol of two weeks [58,59]. Both studies have demonstrated a significant increase in CFI (from 0.116 to 0.159; *p* = 0.028 respectively from 0.21 to 0.31; *p* < 0.05). Of note, this beneficial e ffect of GM-CSF in the promotion of coronary collateral growth could not be transferred to the clinical setting of peripheral vascular disease, where it failed to improve the walking time [60]. Further, one of the clinical trials using GM-CSF for arteriogenesis had to be stopped prematurely for safety concerns in the context of two patients with acute coronary syndrome in the treatment group [59].

G-CSF has been reported in meta-analyses to be safe in terms of major adverse cardiovascular events (cardiovascular death, recurrent myocardial infarction and in-stent restenosis) and toleration of the treatment injections [72–74]. These findings and promising animal test results [75] have led to a randomized, placebo-controlled clinical trial in humans, in which subcutaneous G-CSF was shown to increase CFI from 0.121 to 0.166 (*p* < 0.0001) when administered every other day for two weeks [63]. Despite the above meta-analyses, one study assessing the outcomes and risks of G-CSF in patients with CAD has reported an increased frequency of adverse outcomes (i.e., acute coronary syndrome) [64].

In conclusion, despite the promising results of small clinical trials or animal models using biochemical concepts to therapeutically promote the coronary collateral circulation, none of the approaches evaluated so far could be successfully translated into clinical practice. Besides the above mentioned limitations such as ine fficient collateral formation (i.e., angiogenesis) or potentially harmful propagation of atherogenesis (i.e., the "Janus Phenomenon" [57]), a number of additional unresolved issues remains. These include questions relating to the dosage, the application route and the timing of administration of growth factors [44]. Importantly, considering that "no-option" patients with extensive CAD are the most likely candidates for coronary arteriogenesis, safety of any collateral-promoting substance is crucial [59].

### *4.2. Biophysical Concepts*

The biophysical concept of arteriogenesis is to increase tangential vascular shear stress in preformed coronary anastomoses. One of the natural ways of increasing vascular shear stress is physical exercise [76]. However, because of di fferent comorbidities it is often not feasible for patients with CAD to perform physical exercise training su fficiently. Thus, several other biophysical approaches have been introduced and will be described subsequently.

### 4.2.1. Physical Exercise

The positive e ffects of physical exercise on the cardiovascular system have been known for a long time [77]. For instance, it was concluded in 1958 by Morris and Crawford that physically active people are less prone to develop stable CAD in comparison with sedentary people [78]. Physical exercise has a positive e ffect on several cardiovascular aspects such as vascular remodelling, increase of the maximal coronary blood flow (i.e., coronary flow reserve; CFR) as well as a decrease of coronary artherogenesis [79,80].

Concerning the e ffect of training on coronary collateral function, Scheel et al. observed an arteriogenetic e ffect of physical exercise in dogs with a constricted coronary artery whereas this e ffect was not observed in dogs without coronary stenosis [81]. In the groups with artificial coronary occlusion, exercise stress doubled the collateral growth and hence, the coronary flow reserve when compared to none exercised dogs. In humans, Zbinden et al. documented an increase of the quantitative parameter CFI in a proof-of-concept study [82]. They evaluated CFI, CFR and other cardiac parameters before and immediately after exercise training of a healthy marathon runner and demonstrated an increase of CFI from 0.23 to 0.37. Two small, non-randomized clinical trials have supported the positive e ffect of physical exercise on coronary collateral function, the increase in coronary cross-sectional area [83] as well as dose-dependent relation between training and increase in CFI [84]. Those results are in agreemen<sup>t</sup> with several other studies [80,85–89], which showed augmented perfusion by collateral vessels in response to exercise training. Besides, there have been other clinical trials failing to show a beneficial e ffect on the collateral circulation by exercise as assessed by angiographic imaging, but not by functional measurements [90]. However, the authors mention the limited validity of the angiographic approach and despite the negative outcome on coronary collateral formation, the exercise group had a significant better clinical outcome concerning the frequency of cardiac symptoms and the physical performance.

Recently, the first randomized clinical trial on the e ffect of physical exercise on coronary collateral function has been published [91]. Möbius-Winkler et al. randomly assigned 60 patients to two training groups (moderate- and high-intensity exercise with 10 h of training per week in each group) and one control group (usual care with encouragemen<sup>t</sup> to perform regular physical activity according to current recommendations). After four weeks, both exercise groups showed a significant increase in CFI (from 0.142 to 0.198, *p* = 0.005 respectively from 0.143 to 0.202, *p* = 0.004) without a statistically relevant di fference between the training modalities whereas CFI in the control group remained unchanged (from 0.149 to 0.150, *p* = n.s.).

In conclusion, the positive e ffect of physical exercise on the human coronary collateral circulation has been repeatedly demonstrated. However, there remain important questions concerning the type and extent of physical exercise for optimal promotion as well as the implementation for patients with limited physical possibilities.

### 4.2.2. External Counterpulsation (ECP)

"External counterpulsation therapy was first developed as a resuscitative tool to support the failing heart and was based on the hemodynamic principles of the intra-aortic balloon pump", which is the augmentation of diastolic blood flow with consecutive improvement of coronary perfusion as well as ventricular afterload reduction [92]. ECP uses three pairs of pneumatic cu ffs wrapped around each of the lower extremities. Those cu ffs are sequentially inflated from distal to proximal triggered by the ECG. Besides augmenting diastolic blood flow and reducing ventricular afterload, ECP increases tangential endothelial shear stress triggering arteriogenesis. Used as a safe, e ffective and low-cost second line treatment in refractory angina pectoris, ECP has been shown to be e fficacious in reducing CAD symptoms as well as improving exercise time [92–96]. Of note, the positive e ffect appears to outlast the actual, conventional seven week period of treatment [97].

The e ffects of ECP on the coronary collateral circulation have been evaluated in two invasive clinical trials. Buschmann et al. demonstrated in a non-randomized study a significant increase in CFI. Other invasive parameters obtained in that study as the index of microvascular resistance (IMR) or quantitative coronary angiography (QCA) remained unchanged and hence, the increase of CFI reflected a "true" improvement of the myocardial blood flow [98]. These results have been confirmed in a randomized, sham-controlled clinical trial with an increase in CFI from 0.125 to 0.174 at a four-week follow-up exam (*p* = 0.006) in the experimental, but not in the placebo group (CFI changed from 0.129 to 0.111, *p* = 0.14) [46].

Recently, the principle of external counterpulsation has been individualized in order to alleviate the side e ffects of ECP (i.e., the cumbersome procedure with high pressure levels), thus increasing its acceptance. The so called individual shear rate therapy (ISRT) adjusts the used treatment pressures of the pneumatic cu ffs according to individually adapted intra-arterial shear rates to achieve the same

effect with reduced pressure values [99]. The calculation is based on Doppler-flow parameters in the common carotid artery at di fferent treatment pressure values. Due to this procedure, the individually calculated treatment pressure ranged between 160 to 220 mmHg instead of the regular treatment pressure of 250 to 300 mmHg.

### 4.2.3. Coronary Sinus Reducer

The biophysical concept of the coronary sinus reducer is based on a perioperative approach during heart surgery with artificially narrowed coronary sinus for augmented retro-perfusion [100]. The exact pathophysiologic principle for a beneficial e ffect remains unclear [101]. One proposed mechanism assumes that the venous back pressure as applied in the coronary sinus is regionally balanced in the venous, but not in the vascular bed upstream of the microcirculation [102]. Based on the two regionally counteracting responses of the microcirculation during myocardial ischemia, that is maximal vasodilatation and increased myocardial compressive forces (i.e., augmented ventricular wall stress due to diminished myocardial thickness), regional imbalance in microvascular resistance with higher resistance in the ischemic area arises. Thus, augmented venous back pressure is able to reach the non-ischaemic microcirculation more easily than the ischaemic one, thereby increasing the microcirculatory resistance in the non-ischaemic zone. This leads to a flow diversion of arterialised blood to the ischaemic area at risk under the necessary condition of functional collateral connections originating from the non-ischaemic area [102].

Due to advances in percutaneous coronary intervention, and at the same time, increasing number of patients with refractory angina pectoris, several investigators have picked up this approach by using balloon-expandable, hourglass-shaped devices to physically narrow the coronary sinus [103–105]. In a first-in-man study, this device has demonstrated relief of angina pectoris in 12 out of 14 patients without options for coronary revascularization [103]. Subsequently, Verheye et al. performed a randomized, sham-controlled clinical trial in 104 patients, which confirmed the results of the first study [101]. They showed an improvement in Canadian Cardiovascular Society (CCS) score as well as quality of life. Nevertheless, exercise time and mean change in the wall motion index as assessed by means of dobutamine stress echocardiography remained unchanged [101]. Subsequently and to evaluate the role of this device in future clinical practice, a post marketing study is currently enrolling selected patients without revascularization options (called Reducer-I-study; NCT02710435). They plan to recruit over 400 patients and assess the clinical e fficacy as well as the long-term outcome with follow-ups up to five years after implantation [104].

### 4.2.4. Pharmacologic Biophysical Arteriogenesis

Ivabradine, a specific inhibitor of the If-channel mainly expressed in sinoatrial nodal cells [106], specifically decreases the heart rate without a ffecting cardiac contractility, afterload or vasomotion as it occurs with beta-blockers [107,108]. Based on the biophysical rationale of diastolic prolongation by ivabradine with extension of diastolic vascular shear stress, a small randomized placebo-controlled trial has demonstrated a significant increase in CFI by ivabradine (from 0.107 to 0.152, *p* = 0.0461) [109]. This result is in accordance with several other trials, which have shown an arteriogenic e ffect on coronary arteries by initiating bradycardia [110–112]). However, despite the promotion of coronary collateral supply, ivabradine has not been shown to be e fficacious with respect to cardiovascular outcomes (composite of death from cardiovascular causes or nonfatal myocardial infarction) in patients with stable CAD [106] in bigger randomized trials such as the BEAUTIFUL [113] and SIGNIFY trials [114].

### **5. Therapeutic Promotion of Extracardiac Coronary Supply**

The anatomical connection between the IMAs and the coronary arteries via their most proximal branch (i.e., the pericardiacophrenic artery departing at the first or second intercostal space) is well documented [21,22,25] (Figures 1 and 2). Additionally, due to the connection of the IMAs with the iliac external arteries via the superior and inferior epigastric arteries, collateral supply from the caudal side amounts to approximately two thirds of the flow during IMA patency [102]. This dual blood supply along with the direct anatomical connection to the coronary circulation provided the rationale for the IMA ligation method as a surgical treatment for angina pectoris. Using a small incision between the second and third rib under local anesthesia, transthoracic surgical access and ligation of the IMAs was first performed by D. Fieschi in 1939 (i.e., before the advent of modern cardiac surgery with cardioplegia and heart-lung-bypass) [115]. Later on, the approach was tested by a series of trials carried out in the late 1950s [116–122]. The primary end point of those clinical trials was angina pectoris and, inconsistently, ECG signs of myocardial ischemia. Battezzati et al. [116], after identifying anew a connection between both IMAs and the myocardium, reported consistent improvement in terms of cardiac symptoms in a uncontrolled trial among 304 CAD patients in 1959. Notable, this improvement was sustained during a follow-up of up to four years after the surgical intervention. In a further uncontrolled trial among 50 CAD patients, Kitchell et al. [123] reported similarly favorable results with symptomatic relief in 68% of the patients undergoing bilateral IMA ligation.

The following sham-controlled trials of bilateral IMA ligation in 35 CAD patients coined the phrase "surgery as placebo" in the context of their negative results [124–126]. Although the introduction of a sham-control study design in the context of surgical trials was seminal [127], the conclusion drawn from the negative results of the IMA ligation trials at hand is questionable. In the trial by Cobb et al. [125], angina pectoris relief was found in five of eight patients (63%) after IMA ligation and in five of seven patients (71%) after IMA sham ligation. Dimond et al. [126] reported nine of 13 patients in the verum and five of five patients in the sham-operation group, respectively. Thus, the abrupt stop of bilateral transthoracic IMA ligation was mainly caused by the advent of modern cardiac surgery with bypass grafting rather than by the slim evidence against IMA ligation claimed by the controlled trials. Especially because the soft study end point of angina pectoris would have required patient numbers at one order of magnitude higher than those recruited for the sham-controlled IMA ligation trials [125,126].

Because of the slim evidence against IMA ligation and the promising surgical results in terms of symptomatic relief, this therapeutic concept was revived 75 years after the first attempt using percutaneous interventional techniques. In the context of soft study endpoints, the first observational interventional study on the function of coronary supply by the IMAs has predefined intracoronary ECG (i.c.ECG) ST-segment elevation during coronary occlusion, not angina pectoris, as the first end point for ischemia [128]. In this trial, myocardial ischemia has been induced twice with and without simultaneous IMA occlusion by proximal coronary balloon occlusion in the process of CFI measurement. Further, to eliminate the e ffect of coronary collateral recruitment or ischemic preconditioning occurring during the second (but not the first) occlusion on the collateral circulation, CFI measurement with simultaneous IMA occlusion was performed before the control measurement without IMA occlusion. Despite this conservative study design, the approach showed a consistently reduced i.c.ECG ST-segment elevation during ipsilateral IMA with RCA or LAD occlusion as an expression of reduced ischemia. Further, CFI has been found higher in the presence versus the absence of IMA occlusion in 68% of the measurements, and overall, this di fference amounted to +0.025 compared with the absence of IMA occlusion (*p* < 0.0001) [128]. However, contralateral IMA occlusion did not cause an e ffect indicating the necessity of anatomic vicinity. In this trial, functional connection between the coronary arteries and the IMAs was slightly less frequent in case of LAD with left IMA occlusion (25 of 30 measurements) than in the case of RCA with right IMA occlusion (28 of 30 measurements).

Based on those functional findings, an anti-ischemic therapeutic approach consisting in distal IMA occlusion by interventional techniques could be a promising therapeutic alternative to IMA bypass grafting. In an open-label proof-of-concept study, Stoller et al. investigated a catheter-based permanent IMA occlusion in the setting of the less frequently grafted right IMA among patients with ischemia in the RCA territory [129]. In this study, 50 patients with chronic stable CAD underwent permanent device occlusion of the distal right IMA. CFI of the RCA measured immediately before and six weeks after the IMA-occlusion showed a consistent increase from 0.071 at baseline to 0.132

(*p* < 0.0001). Further, this augmented coronary blood supply was reflected by the i.c.ECG as a direct measure of myocardial physiology revealing a decreased ischemia during RCA occlusion from baseline to follow-up examination (*p* = 0.0015). Figure 5 illustrates this increased collateral function along with decreased myocardial ischemia during coronary occlusion as outlined by an absent ST-deprivation in the ECG of the follow-up intervention.

**Figure 5.** Collateral flow index (CFI) measurements of the right coronary artery (RCA) with corresponding electrocardiograms (ECG) after a one-minute proximal coronary balloon occlusion. (**A**) CFI measured immediately before permanent right internal mammary artery occlusion showing a collateral blood supply of 0.100 and marked ST-deprivations in the ECG as a sign of ischemia (marked with an arrow). (**B**) Six weeks after the permanent occlusion, CFI increased to 0.250 (+0.150). This augmented coronary blood supply is reflected by the ECG revealing a decreased ischemia without ST-deprivations (marked with an arrow).

To conclude, augmentation of extracardiac coronary supply by permanent right IMA device occlusion is effective and feasible. However, if and how this increased collateral blood flow improves clinical outcome parameters is subject of current research. For this reason, a randomized, sham-controlled and double-blind clinical trial is currently enrolling patients (NCT03710070). It aims to include 250 patients in order to assess the clinical efficacy (measured as treadmill exercise time increment) in the next few years.
