Coronary flow reserve evaluated by phase-contrast cine cardiovascular magnetic resonance imaging of coronary sinus: a meta-analysis

We searched the electronic databases PubMed, Web of Science Core Collection, Cochrane Advanced Search, and EMBASE using the search formulas listed in the Appendix (Additional file 1). We performed the database search on May 25, 2022. After screening all abstracts from the search results, potentially relevant studies were reviewed by two reviewers (SK and MA) for final eligibility. A third reviewer resolved disagreements between them. This meta-analysis adhered to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. This protocol was registered in the University Medical Informatics Network (R000054826). We did not obtain institutional review board approval since this study was a meta-analysis.

We included prospective and retrospective studies that included data on CFR evaluated by phase-contrast CMR of the coronary sinus in patients with cardiovascular diseases and CAD risk factors. The subjects included healthy individuals and controls (a group of patients who underwent CMR for some clinical reason but in whom no abnormalities were detected), those with CAD, and those with cardiomyopathy (hypertrophic cardiomyopathy [HCM] and dilated cardiomyopathy [DCM]). Studies of other diseases were excluded from the meta-analysis. We did not include studies on myocardial perfusion reserve using stress-perfusion CMR, as the methodology is completely different. Only articles published in English were included in this study.

Two reviewers extracted the study characteristics, including author name, publication year, country of origin, patients’ diseases, and CMR parameters (SK, MA). First, the disease, age, sex, and CMR parameters for each study were extracted (in addition to CFR from phase-contrast cine CMR, which was the focus of the study, MBF calculated from blood flow in the coronary sinus, left ventricular (LV) function, LV volume, and late gadolinium enhancement (LGE). Second, we compared the CFR between cardiovascular disorders and healthy controls. Finally, the hazard ratio (HR) of CFR for the major adverse cardiac event (MACE) was evaluated in patients with CAD.

In the literature, two separate methods are used to calculate CFR. The first method calculates the CFR by dividing the blood flow in the coronary sinus at stress by the blood flow at rest. The second method is to divide the coronary sinus blood flow by the LV myocardial weight to calculate the MBF (mL/min/g) and then divide the MBF at stress by the MBF at rest to calculate CFR. Because the CFR values are the same for both methods, those calculated by these two methods are harmonized. We also used the Newcastle–Ottawa Quality Assessment Scale and Case Control Studies to evaluate the risk of bias [8] (Additional file 1).

A random model meta-analysis was performed using RevMan 5.41 (Cochrane Collaboration, London, UK). Using the mean difference (MD), the CFR results were compared between patients with cardiovascular disease (stable CAD, HCM, DCM), and a healthy control HR meta-analysis was performed of stable CAD using the general inverse variance method. Heterogeneity was indicated by I², where 0% indicated no heterogeneity and 100% indicated strong heterogeneity (p < 0.05), which was considered statistically significant.

Of the 565 candidate studies, 47 were deemed eligible for inclusion [7, 9‐55] (Fig. 1). Fourteen studies presented two populations each [9, 11, 15‐17, 23, 25, 27, 29, 31, 32, 34, 47, 49], while two studies presented four populations each [30, 35]. Therefore, a total of 67 independent populations were included. Among the 47 studies, 22 were from Japan [19‐21, 23‐37, 42, 44, 45, 47], eight from the United States [7, 14, 22, 40, 41, 48, 51, 55], five from Sweden [12, 16, 17, 53, 54], three from Germany and Sweden [11, 15, 46], two from Finland [38, 39], and one each from Australia [18], France [43], Italy [9], Norway [13], Turkey [10], the United Kingdom [49], Switzerland [7], and the Netherlands [50]. Year of publication ranged from 1992 to 2022. In 35 studies, 1.5T CMR systems were used [7, 9‐12, 15‐17, 19‐21, 24‐32, 34‐40, 42, 46‐49, 51, 52, 54]; in nine studies, 3T CMR systems were used [13, 14, 18, 22, 33, 41, 43, 53, 55]; in one study, both 1.5 and 3T CMR units were used [44]; and in one study, a 0.6T CMR system was used [50]. Information on pharmacological stress was extracted from 41 studies. Four studies used the cold pressor test [21, 33, 41, 43], while two studies did not perform any stress testing [9, 50]. Because CPT detects endothelial function–dependent increases in blood flow that differ significantly from pharmacological stress, only resting blood flow was extracted and integrated from the four CPT studies [21, 33, 41, 43]. A total of 28 (68%) studies used adenosine triphosphate infusion, 10 (24%) used dipyridamole, and three (8%) used the regadenoson. However, no study has employed exercise-stress testing. The median value of velocity encoding (VENC) was set at 50 cm/s at rest (range, 40–200 cm/s). Most studies used the same VENC during rest and stress. In six studies, different VENC were used during pharmacological stress. In four studies [10, 12, 40, 48], the VENC during stress was 200 cm/s. In the study by Gyllenhammar et al. [16], the VENC was 120 cm/s. According to Moro et al. [43], the VENC was 150 cm/s. Thirteen studies performed phase-offset correction using adjacent myocardial tissue [21, 28‐31, 33, 35‐39, 47, 55], and one study used static tissue regions in the chest wall [55]. The standard imaging parameters and analytical images related to phase-contrast cine CMR are presented (Additional files 1, 2).

Among the control subjects, the pooled CFR from the meta-analysis was 3.48 (range, 2.98–3.98; I² = 97%; p for heterogeneity < 0.001) (Fig. 2). The estimated CFR was 2.50 (range, 2.38–2.61; I² = 97%; p for heterogeneity < 0.001) for patients with stable CAD (Fig. 3) and 2.01 (range, 1.70–2.32; I² = 82%; p for heterogeneity < 0.001) for those with cardiomyopathies (HCM and DCM) (Fig. 4). Coronary sinus blood flow and MBF values are summarized in Figs. 5, 6 and 7. Among the control subjects, mean coronary sinus blood flow was 100 mL/min (range, 83–118 mL/min) at rest and 312 mL/min (range, 251–374 mL/min) during stress, while the mean MBF was 0.79 mL/min/g (range, 0.69–0.90 mL/min/g) at rest and 2.8 mL/min/g (range, 2.13–3.39 mL/min/g) during stress (Fig. 5).

Among the stable CAD patients, mean coronary sinus blood flow was 102 mL/min (range, 94–110 mL/min) at rest and 240 mL/min (range, 222–257 mL/min) during stress, while the MBF was 0.9 mL/min/g (range, 0.8–1.0 mL/min/g) at rest and 2.0 mL/min/g (range, 1.8–2.2 mL/min/g) during stress (Fig. 6).

Among the patients with cardiomyopathies, the mean coronary sinus blood flow was 127 mL/min (range, 99–155 mL/min) at rest and 253 mL/min (range, 171–336 mL/min) during stress, while the mean MBF was 0.7 mL/min/g (range, 0.5–0.8 mL/min/g) at rest and 1.8 mL/min/g (range, 1.0–2.7 mL/min/g) during stress (Fig. 7). The meta-analysis showed that the mean CFR was significantly reduced among the patients with stable CAD (MD = -1.48; 95% CI, -1.78 to -1.17; p < 0.001; I² = 0%; p for heterogeneity = 0.48) (Fig. 8A), HCM (MD = -1.20; 95% CI, − 1.63 to − 0.77; p < 0.001; I² = 0%; p for heterogeneity = 0.49) (Fig. 8B), and DCM (MD = -1.53; 95% CI, − 1.93 to − 1.13; p < 0.001; I² = 0%; p for heterogeneity = 0.45) (Fig. 8C) compared to the control subjects.

Four studies showed outcomes data for patients with stable CAD [22, 27, 30, 35] (Table 1). All studies defined the outcome as a composite endpoint (MACE). Indorkar et al. [22] defined MACE as death, nonfatal myocardial infarction (MI), heart failure hospitalization, sustained ventricular tachycardia, and late revascularization. Kanaji et al. [25] defined MACE as all-cause death, nonfatal MI, hospitalization for congestive heart failure, and stroke. In another study, Kanaji et al. [27] defined MACE as cardiac death, MI, clinically driven remote revascularization (> 3 months after the index percutaneous coronary intervention), or hospitalization for heart failure (HF). Kato et al. [35] defined MACE as cardiovascular death, acute MI, unstable angina, HF hospitalization, or ventricular tachyarrhythmia necessitating defibrillation. A meta-analysis revealed that a higher CMR-derived CFR was significantly associated with a lower MACE rate (HR = 0.52 per unit increase; 95% CI, 0.37–0.73; p < 0.001; I² = 84%; p for heterogeneity < 0.001) (Fig. 9).

The epicardial coronary arteries (diameter > 400 μm) are conductance vessels, whereas the pre-arterioles (diameter 100–400 μm) and intramural arterioles (diameter < 100 μm) are resistance vessels [5]. Myocardial flow distribution is mainly regulated by small vessels such as the pre-arterioles or arterioles. In MVD, several structural and functional abnormalities occur in these small vessels, such as microvascular remodeling and endothelial dysfunction [56]. These abnormalities are related to attenuated vasodilator response of the smooth muscle cells of vessels resistive to adenosine and dipyridamole associated with impaired CFR. CFR is a functional measure of the epicardial coronary artery and the coronary microcirculation. In the absence of epicardial stenotic disease, a decreased CFR can be a surrogate marker of coronary microvascular dysfunction [57]. PET is the most firmly established non-invasive quantitative assessment method for MBF and CFR. Recent data have demonstrated that the prevalence of a CFR impairment is higher than expected in patients with known or suspected CAD [5]. Moreover, the predictive value of PET-derived CFR is well established [6]. Based on robust clinical evidence, the United States Food and Drug Administration approved the application of ⁸²Rubidium and ¹³N-Ammonia PET for assessing CFR [5]. However, the utility of PET for evaluating myocardial perfusion has not yet been established.

Despite the rich clinical evidence of PET imaging for assessing CFR, it features some limitations, such as radiation exposure, limited accessibility, and high cost. Phase-contrast cine CMR of the coronary sinus is also used to quantify CFR. This method has an absolute advantage over PET in that it does not involve radiation exposure. The coronary sinus theoretically drains approximately 96% of the total LV myocardial blood flow [58]; therefore, the LV myocardial blood flow can be calculated by measurement of coronary sinus blood flow. Previous studies reported that phase-contrast cine CMR-derived coronary sinus blood flow correlates well with myocardial blood flow by PET [7]. This evidence demonstrates the high accuracy of CMR-derived CFR.

To date, clinical evidence of CMR-derived CFR is relatively limited. Therefore, here we performed a meta-analysis to determine its clinical relevance. Compared with controls, we found a significant decrease in CFR in various cardiovascular diseases, such as CAD and cardiomyopathies (Figs. 2, 3 and 4). Moreover, CMR-derived CFR can provide prognostic information for patients with CAD. Regarding its technical aspects, setting VENC and phase-offset correction is essential to an accurate evaluation of the blood flow within the coronary sinus. The median VENC was 50 cm/s at rest (40–200 cm/s). Suboptimal VENC settings can lead to aliasing, which can lead to blood flow under- or overestimations. A 40 cm/s VENC at adenosine loading is generally not considered sufficient to increase coronary flow by a factor of 3 to 4, especially during stress; therefore, the VENC should be increased appropriately.

Regarding phase offset correction, 13 studies used adjacent myocardial tissue and one used static tissue regions at the chest wall. Other studies did not perform phase offset correction. Some studies reported the importance of the impact of phase offset correction but did not explain why phase offset correction was not performed. In addition, because of our limited technical knowledge of phase offset correction, we could not adequately determine its importance (whether it is absolutely necessary) in this study. Further research on this topic is required. Evaluation of the effect of medical therapy, such as angiotensin-converting enzyme inhibitors/angiotensin receptor blockers and statins on CFR, is of interest. A recent meta-analysis showed the impact of medical therapies on CFR [59]. However, CMR data were not included thus, further studies are warranted to assess the utility of phase-contrast CMR in the serial impact of medical therapies on CFR.

We believe that our data will contribute to the accurate assessment of CFR using phase-contrast cine CMR of the coronary sinus. However, the relationship between the CMR-derived CFR and stress-perfusion CMR should be recognized. Although this study shows great potential for CMR-derived CFR, the mainstream method remains stress-perfusion CMR. CMR-derived CFR should be performed within a protocol featuring stress-perfusion CMR. However, CMR-CFR has the potential to assess the global microcirculation, which is difficult to assess using stress-perfusion CMR, and may add new CMR value to the diagnosis of INOCA in the post-ischemia era [60].