Cardiovascular consequences of myocardial bridging: A meta-analysis and meta-regression

Myocardial bridging, a congenital abnormality in which a coronary artery tunnels through the myocardial fibres was usually considered a benign condition. Many studies suggested a potential hemodynamic significance of myocardial bridging and some, usually case reports, implied a possible correlation between it and various cardiovascular pathologies like acute myocardial infarction, ventricular rupture, life-threatening arrhythmias, hypertrophic cardiomyopathy, apical ballooning syndrome or sudden death. The main objective of this article is to evaluate whether myocardial bridging may be associated with significant cardiac effects or if it is strictly a benign anatomical variation. To this purpose, we performed a meta-analysis (performed using the inverse variance heterogeneity model) and meta-regression, on scientific articles selected from three main databases (Scopus, Web of Science, Pubmed). The study included 21 articles. MB was associated with major adverse cardiac events - OR = 1.52 (1.01–2.30), and myocardial ischemia OR = 3.00 (1.02–8.82) but not with acute myocardial infarction, cardiovascular death, ischemia identified using imaging techniques, or positive exercise stress testing. Overall, myocardial bridging may have significant cardiovascular consequences (MACE, myocardial ischemia). More studies are needed to reveal/refute a clear association with MI, sudden death or other cardiovascular pathologies.

Search method. We analyzed the results obtained from three databases: Web of Science, Scopus, and Pubmed, by using the following keywords: "myocardial bridging", "mural coronary". We preferred not to use additional, restrictive criteria (e.g. article type) as other varieties (letters, case presentations, reviews) could add relevant data to the meta-analysis (discussions, finding other pertinent studies). The reference list of each appropriate item was inspected for other, potentially relevant studies, to be included in the meta-analysis. The references and abstracts (if available) were imported in Thompson Endnote software, and useful articles were downloaded for further inspection.
Data collection and analysis. For each study, two reviewers extracted the data separately and included it in Excel Datasheets. We summarised the following information: study, name, the total number of cases in each group, country, general inclusion and exclusion criteria, endpoints, covariates: depth, length, systemic hypertension, angina pectoris, diabetes mellitus, hyperlipidemia, smoking, age, gender.
Risk of bias. The risk of bias was assessed separately, for each case, by two reviewers. We analysed selection bias (inclusion and exclusion criteria, whether the study was dedicated to MB or had a more general nature), multiple publication biases, measurement bias (the method used), statistical reporting bias. Based on these elements, we separated the studies in three subgroups: high risk of bias, moderate risk of bias and low risk of bias. Quality assessment. Quality assessment was performed with the Quality in Prognostic Studies Tool from which we removed study attrition (which was considered not relevant for a prevalence study). For each remaining subscales (study participation, prognostic factor measurement, outcome measurement, study confounding, statistical analysis and reporting), we graded each study as high quality (2 points), moderate quality (1 point) or low quality (0 points) and we summed up the results (Table 1).

Statistical analysis.
We determined the effect size using inverse variance heterogeneity model computed in Microsoft Excel 2016 with the MetaXL add-on version 5.3 and used Comprehensive Meta-Analysis v3 for performing meta-regression. For each group and subgroup, we performed a forest plot. For the analysis of publication bias, we used the funnel plot and the Luis Furuya-Kanamori (LFK) index. We used a continuity correction of 0.5, and 95% confidence intervals. Forest plots were done using Microsoft Excel 2016 with the MetaXL add-on 5.3, with the effect size transformed logarithmically for better viewing.

Results
Search synthesis. During the initial database research, we obtained 7358 articles from which, after deleting 4839 duplicates, we were left with 2519, that were included in the initial analysis. From them, we screened the abstracts and removed all articles not containing clinical studies that included a group of subjects with myocardial bridging and a group without this condition, regarding the above-mentioned endpoints. 32 articles selected and further downloaded. By analysing their references, we found another 5 potentially relevant items, which were also downloaded. The next step included the analysis of the full text manuscripts, after which we selected 21 to be included in the meta-analysis, which fully respected the inclusion criteria. Details about search synthesis are presented in Fig. 1 40 . We detailed the papers comprised in the analysis in Table 1.

Major Adverse Coronary Events (MACE).
Sixteen studies contained data about a potential correlation between MB and MACE. The presence of MB led to a significantly increased risk for MACE -OR = 1.53(1.01-2.30), p < 0.001 (Fig. 2). The publication bias was minimal, with an LFK index of 0.33. The heterogeneity of the included studies was moderate (I 2 = 61%). Meta-regression analysis found a positive correlation between MACE risk and the increase in age differential between subjects with MB versus subjects without MB; gender, systemic hypertension, hyperlipidemia, previous angina, diabetes mellitus or smoking failed to generate statistically significant differences (see Table 2). By performing a subgroup analysis, based on whether the studies considered significant coronary artery disease as an either inclusion or exclusion criteria, we found that, if the presence of CAD was used as an inclusion criteria, OR was 1.60(0. 79 (Fig. 4). The publication bias was minimal (LFK index = 0.30). The heterogeneity of the included studies was low (I 2 = 27%). Meta-regression analysis did not show any significant effects on the result determined by age, gender, systemic hypertension, hyperlipidemia, previous angina, diabetes mellitus or smoking (see Table 2). By performing a subgroup analysis, based on whether the studies considered significant coronary artery disease as an either inclusion or exclusion criteria, we found that, if the presence of CAD was used as an inclusion criteria, OR for MI was 0.74(0. 43-1.25), and if it was used as an exclusion criteria -1.45 (0.87-2.39).

Cardiovascular death.
Eight studies included data about a potential correlation between MB and cardiovascular death. The presence of MB did not lead to a significant increase in the risk of cardiovascular death -OR = 2.19(0.65-7.43) (Fig. 5). The publication bias is minor (LFK index = −1.28). The heterogeneity of the included studies was moderate to high (I 2 = 70%). Meta-regression analysis found a positive correlation between CVD risk and the increase in age differential between subjects with MB versus subjects without MB; (see Table 2).

Myocardial ischemia.
Nine studies included data about a potential correlation between MB and myocardial ischemia. Two studies contained more than one method of assessing myocardial ischemia (Xiang and Schwartz), and we took into account in both cases the outcome with the most number of positive subjects. The presence of MB was significantly associated with myocardial ischemia -OR = 3.00(1.02-8.82) (Fig. 6). The publication bias was minor (LFK index = −1.68). The heterogeneity of the included studies was high (I 2 = 89%). Meta-regression analysis did not show any significant effects on the result determined by age, gender, systemic hypertension, hyperlipidemia, previous angina, diabetes mellitus or smoking (see Table 2).

Discussions
Our study revealed that, overall, MB increases the risk for MACE and myocardial ischemia, and subsequently a possible connection between it and various cardiovascular pathologies cannot be excluded. The physiological characteristics and morphological consequences of MB are still not yet fully elucidated, even if there are many studies analysing specifically with these issues. Even if MB mainly causes a systolic narrowing of the affected coronary, both cardiac phases are affected. Intracoronary Doppler showed a finger-tip-like anterograde flow of the blood in most MB patients, characterised by a sharp increase in coronary flow velocity during early diastole, followed by a steep decrease in speed, leading to a plateau in mid-late diastole 41 . During systole, the anterograde flow is decreased, absent, or even reversed, due to the systolic squeezing of the bridged coronary segment 41 . Ge et al. showed the pressure within the segment proximal of the MB to be much higher than the pressure in the aorta and that the maximum pressure in the proximal segment is reached in mid-systole compared to normal segments is reached in the end-systole 42 . During end-diastole, the myocardium dilates, triggering a negative pressure in the mural segment, and an increased pressure gradient between the bridged segment and the neighbouring areas(the so-called sucking phenomenon) 42 . The systolic compression is usually eccentric, and only rarely concentric, leading to a half-moon like echo-lucent area neighbouring the bridge when viewed through IVUS 41 . Previous studies showed that systolic compression by the bridge could be either elliptical, associated with a 50% reduction of the luminal area, or concentric, associated with a 75% reduction 43 . Other authors showed that, if superficial and short MBs do not lead to significant hemodynamic effects, this is not the case with deep and long variants, which may cause severe ischemia or cardiovascular events 44 . Hostiuc et al. showed that hemodynamically significant MB is associated with increased fibrosis and interstitial edema in subjects who died suddenly and had no other apparent cause of death, suggesting that the most likely cause of death was electrical instability 6 , which is known to be associated with increased myocardial fibrosis 45,46 .
Endothelial dysfunction is an early phase of atherogenesis 47 . Numerous studies reported an alteration of the normal endothelial physiology associated with MB, including an abnormal response to acetylcholine (ACh) or decreased responses to NO and endothelin-1. Acetylcholine acts as an endothelium-dependent vasodilator, by stimulating the release of EDRF, if the endothelium is intact 48 . If however there are endothelial lesions (including atherosclerotic disease), ACh will cause smooth muscle constriction 49 . ACh-generated vasoconstriction was often showed to be associated with MB. For example, Kuhn revealed that vasoconstriction after the administration of ACh is higher in the coronary segment containing a MB compared to non-bridged segments 50 . Kim et al. showed that, after infusion with ACh, subjects with MB have a statistically significant increase in focal vasoconstriction in the mural segment, compared to similar, non-bridged coronary segments from the control group 51 . Teragawa showed that subjects with MB have an almost two-fold increase in the rate of coronary spasm compared with the control group 52 . Zoghi et al. showed, by measuring flow-mediated dilatation in the brachial artery, that the presence of MB also leads to remote vasoconstrictive effects 53 . Herrmann demonstrated that MB causes an increased shear stress and significant vasoconstriction after the administration of ACh without changes in the coronary blood flow 16 . Masuda et al. found that bridged LAD segments have a weaker immunoreaction to endothelial nitric oxide synthase (eNOS) and endothelin-1(ET-1) compared to the proximal and distal segments 54 . NO leads to vasodilatation, and it might be involved in the "sucking phenomenon" described by Ge 42 .
The segment proximal to the MB is put at a high wall shear stress due to the significant pressure gradients between the bridged area and the proximal coronary segment 42 , potentially causing a chronic coronary pressure overload (CCPO) 16 . CCPO is associated with significant endothelial changes, including intimal thickening, medial thickening, disruption of the internal elastic lamina or endothelium 55 , favouring the appearance of atherosclerosis.
Nakaura et al. showed that mid-LAD MB is strongly positively associated with atherosclerosis (OR = 4.99). They did not mention, however, the location of the atherosclerotic plaques (distally, intramural or proximal from the bridged area). Hong et al. found that the prevalence of lesions located distally from the MB was significantly lower compared to those located proximally (5.9% versus 62.4%). In their article age, male gender, a history of diabetes mellitus, systemic hypertension, bifurcated vascular lesions, smoking index, or hyperlipidemia increased the severity of the coronary stenosis 56 . Wang et al. showed that MB (both superficial and deep) is associated with non-significant stenosis proximal to the bridged area and that MB in middle LAD is positively associated with stenosis in non-bridged arteries. The most likely explanation for the latter is the fact that deep MB has hemodynamic significance, causing a disturbed flow in the other coronaries, and therefore increases the frequency of atherosclerotic deposits 57 . By compressing the tunnelled artery during systole, MB increases lymph drainage of the vessel wall, protecting against lipid accumulation within the bridged coronary wall 57,58 . Verhagen et al. suggested that the reason for the lack of atherosclerosis under the bridged coronary is represented by a lack of exposure to perivascular adipose tissue at that site, which is known to secrete pro-inflammatory adipokines and cytokines 59 . Saidi et al., on the other hand, showed that between the bridged muscle and the coronary is sometimes interposed an adipose layer, which they hypothesised to act as a cushion, limiting the compressing forces acting upon the bridged coronary segment 60 . However, this layer would, in fact, secrete pro-inflammatory adipokines and cytokines. Locally increased eNOS is known to be associated with increased atherosclerosis in coronary arteries 54,61 ; therefore, increased eNOS levels proximally from the bridged segment might be involved in the development of proximal segment atherosclerosis. Another explanation for the distribution of atherosclerosis/around    the bridged area might be represented by the shape of the endothelial cells, which are spindle-shaped under the bridge, suggesting they are under a high shear stress, while in other regions they have a polygonal distribution, suggesting they are subjected to a decreased shear stress 58,62 . Increased shear stress seems to decrease monocyte adhesion and lipid transfer across the vascular wall, therefore decreasing the risk for atherosclerosis 59,62 . Bilen et al. showed that subjects with MB have a significantly increased mean platelet volume (MPV). Larger platelets are associated with increased reactivity and production of more prothrombotic factors 63 . Potential causes for this result might be a reduced production of prostacyclin and nitric oxide, impairment of   endothelial-dependent vasorelaxation, increased proximal atherosclerosis, increased shear stress 64 . This increased thrombogenic status might favour the development of coronary thrombosis or might increase its severity.
Numerous studies suggested or proved that MB is associated with various types of physiological changes. Ge et al. showed that MB leads to a decreased coronary flow velocity reserve, which remains higher proximally compared to distally from the bridged segment; also, coronary flow velocity reserve seems to be more affected in subjects with severe coronary stenosis 41 . Schwartz et al. quantified the decrease in the coronary flow reserve caused by MB -they found a mean ratio of 2.0 distally from the bridged coronary artery (normally the value is above 3) 7 . Fractional flow reserve is often used to quantify the physiological significance of CAD. It is not influenced by heart rate or blood pressure changes 65 , nor by dobutamine challenge in subjects with fixed CAD 66 . Diefenbach et al. showed that inotropic challenge leads to a significant increase in the degree and length of MB segments, as evaluated on CAG, compared to those obtained at rest in symptomatic subjects, suggesting the presence of a specific coronary physiological response to inotropic stimuli 67 . Yoshino et al. proved that fractional flow reserve (FFR) to dobutamine challenge is significantly decreased in the presence of microvascular endothelial dysfunction 65 . Heart rate recovery is a prognostic tool for the cardiac autonomic functions 68 ; increased heart rate recovery (HHR) was associated with decreased parasympathetic activity 69 , a higher susceptibility for atherosclerosis, and increased five-year mortality 70 . Okutucu showed the presence of MB on LAD to be a significant predictor for increased HHR at 1 minute (b = −8.524; 95% CI −14.934 to −2.113; p = 0.009), and argued that the finding was caused by silent or symptomatic myocardial ischemia. QT dispersion (QTd), defined as the minimum QT interval subtracted from the maximum, is an effective marker for electrical myocardial heterogeneity; higher values are associated with an increased risk of cardiac arrhythmias 71 . Aksan et al. demonstrated that subjects with MB have a significantly increased QTd during peak exercise compared with its baseline value (42.6 ms versus 36.4 ms), while in non-MB subjects the increase during peak exercise was not statistically significant (29.1 ms, versus a baseline value of 27.1 ms) 72 .
All these results come to confirm that, at least in some cases, MB may have various cardiovascular consequences. Our study showed that MB presence is associated with an increased risk of MACE and myocardial ischemia, but failed to show a possible positive correlation with CMH, MI, or cardiovascular death. However, even if these associations did not reach statistical significance, the variability of the confidence intervals is very high, which can be caused either by the heterogeneity of the studies included in the meta-analysis, or by a highly variable response to this condition (possibly dependent upon length, depth or the MB, its location, associated pathologies, and so on). The scientific literature analysis showed or suggested a series of associations between MB and cardiovascular pathologies, including, coronary spasm, MI, HCM, Takotsubo cardiomyopathy, arrhythmias, or sudden death.
The association between coronary artery spasm and MB was first implied by Grover et al., who reported the case of a patient who developed CAS during cardiac pacing, an association that was later on confirmed by other studies (see e.g. ref. 73 ). Recent articles showed that patients with MB have a higher incidence of coronary vasospasm induced by ACh. Sung et al. showed that patients with MB who responded to lower doses of ACh had a more severe decrease of the luminal diameter, and had a worse cumulative clinical outcome after 12 months compared to subjects who needed larger ACh doses for a positive response. The mechanical stress caused by the systolic narrowing is known to lead to endothelial dysfunction. Subjects with variant angina are known to be hyperactive to vasoconstrictor stimuli 74 . Atherosclerotic epicardial vessels may present a paradoxical vasoconstriction caused by a direct stimulation of the mural smooth muscle, as a result of widespread endothelial dysfunction 49 . Corroborated, these results might suggest that the coronary vasospasm associated with MB might lead to the development of myocardial ischemia, especially in the presence of other coronary pathologies 71,75 .
De Agustin et al. showed that subjects with chest pain and without severe coronary artery stenosis are significantly more likely to have MB compared to those with severe CAD; moreover, in the MB subgroup hyperlipidemia was considerably less often found compared to the group of subjects without MB 76 . Tang et al. showed that, in subjects with MB, cardiac ischemia is strongly associated with the degree of mean systolic narrowing, and suggested that the degree of systolic occlusion by the MB is responsible for the appearance of myocardial ischemia 77 . Angelini et al. analysed the presence of MB in subjects with and without coronary artery disease, finding that its incidence is significantly lower in the group without CAD 43 . They also showed that there is a higher incidence of MB in men and in subjects with a systolic overload of the left ventricle.
There are two core mechanisms through which MB is thought to cause myocardial ischemia and MI -the development of atherosclerosis proximally from the bridged segment 78 , and a direct compression of the coronary artery by the MB 79 . In subjects with MI, the thickness of the MB is, on average, greater compared to the ones found in subjects without MI 80 . Takamura et al. studied two groups of subjects: one with MI in the area supplied by the LAD and without MB on LAD and one with MI in any myocardial area associated with MB on LAD, and found that, if the acute occlusion occurred in the LAD segment proximal from the MB, the length and thickness of the MB were significantly greater, and the distance from the ostium of the LCA to the MB was significantly shorter compared to those without an acute occlusion on the LAD proximally to the MB 81 .
The blood flow during systole is small compared to the one from diastole in normal conditions; this is not the case in tachycardia, when the end-diastolic volume decrease is greater compared to end-systolic volume 82 , potentially leading to myocardial ischemia. In this instance, the blood supply of the myocardium becomes more dependent upon the systolic blood flow, which is impaired by the bridge. There were some cases published in the scientific literature in which tachycardia was associated with MB. Feld, for example, described a patient who, during stage 5 of the Bruce protocol, developed significant flat ST depressions, followed by a complex tachycardia with a rate of 280 bpm, associated with a MB on LAD 20 83 . Left ventricular systolic dyssynchrony or intraventricular dyssynchrony, is characterized by a delay or heterogeneity in the timing of contraction in different myocardial segments of the left ventricle. It is usually found in heart failure, and it is proven to cause a more severe myocardial disease and a poorer cardiovascular prognosis. Cai Wei et al. showed that mid-LAD MB is associated with an increased prevalence of left ventricular systolic dyssynchrony (LVSD) compared to subjects without MB; moreover, the stenosis caused by the MB and the length of the mural segment have a synergistic effect on the development of LVSD 84 .
Lee et al. found that MB is an independent predictor for MACE in patients with eluting drug stents (adjusted OR = 2.897) 75 . Similarly, Tsujita et al. found that subjects undergoing eluting drug stents for coronary artery disease in the proximal part of the LAD, and who had detectable MB distal of the stented lesion, had a significantly decreased minim stent area at the end of the procedure, and a higher target lesion vascularization rate 85 .
Hypertrophic cardiomyopathy, an inherited cardiovascular disease that can lead to SCD, heart failure, or stroke, is believed to be associated very often with MB 86,87 . Earlier studies implied a possible correlation between HCM with MB and SCD in some subgroups of patients. For example, Yetman et al. showed that MB in children with HCM is a risk factor for SCD 28 . However, more recent ones failed to prove a positive correlation. Tian et al. showed that, in autopsy cases, MB is not associated with SCD in HCM 44 . Basso et al. found that patients with HCM who died suddenly do not have a higher prevalence of MB compared to subjects with HCM who died from heart failure or other modes of death; from the analysis of their results, however, the age of the subjects with SCD was much lower (22 years, compared with 54 and 67 in the other two groups), the length and depth were higher, and the proportion of associated CAD much lower 87 . Sorajja et al. showed that there is no statistically significant difference regarding five-year survival between subjects with and without MB 88 . Our study showed that, when we excluded studies in which HCM was used as an inclusion criteria, MB increased the risk for MACE. More studies are however needed in the area before either establishing or excluding a causal link between these two pathologies.
Some studies suggested that MB is more frequent in patients with apical ballooning syndrome 26,89 , but not in midventricular or basal ballooning 89 ; in these cases, MB might influence the development of Takotsubo cardiomyopathy through the generation of microvascular dysfunction and altered sympathetic activity, possibly associated with recurrent LAD segments 89 ; more studies are needed in this area to reach definite conclusions.

Limits.
A very high variability of the results, associated with a limited number of studies for testing associations between MB and various cardiovascular consequences. Many studies did not include comprehensive inclusion and exclusion criteria, or complete descriptive data, limiting the number of subgroup analyses we were able to perform. Due to the presence of highly different methodologies, and inclusion and exclusion criteria, there can be a significant selection bias, which was attempted to be minimized through subgroup analyses and meta-regression.

Conclusions
Overall, myocardial bridging may have significant cardiovascular consequences (MACE, myocardial ischemia). More studies are needed to reveal/refute a clear association with MI, sudden death or other cardiovascular pathologies.