A total of 254 patients who underwent coronary angiography due to suspected angina or ischemic evidence between December 2014 and March 2020 were verified to have CTO in only one major epicardial coronary artery and were considered for inclusion in this study. The study protocol was approved by the Ethics Board of Beijing Anzhen Hospital, Capital Medical University. Patients with acute myocardial infarction within 3 months, patients with coagulation disorders, and patients who refused or were unable to undergo CMR (n = 46) were excluded from the study. Patients were scheduled for CMR within 1 week after coronary angiography. A total of 14 patients with poor CMR images or failed procedures were also excluded, leaving a final total of 194 eligible patients that were enrolled in the study. Figure 1 shows a schematic diagram describing the study cohort. All the patients were treated with optimal medical therapy (OMT) (medical therapy formulated by clinicians in full consideration of risk factor modification and permanent lifestyle changes) (12), regardless if the revascularization procedure succeeded or not. Written informed consent was obtained from all the patients and the study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki.

Coronary angiography was performed by using the standard method. CTO was defined as 100% stenosis with the thrombolysis in myocardial infarction (TIMI) grade 0 flow in a major epicardial coronary artery (≥2.5 mm) for at least 3 months (13). Significant non-CTO coronary disease was diagnosed if there was ≥70% lumen stenosis in a major epicardial coronary artery (≥2.5 mm), except for the left main (LM) artery or ≥50% lumen stenosis in the LM artery. The presence and state of collaterals supplying the totally occluded vessel from the contralateral artery were graded by using the Rentrop classification system (14): Grade 0 referred to no collateral circulation, grade 1 referred to collateral circulation that only supplied the occluded vessel branch, grade 2 referred to collateral circulation that partially supplied the occluded vessel trunk, and grade 3 referred to collateral circulation that completely filled the occluded vessel trunk. Patients were classified as having poorly-developed collaterals (Rentrop scores of 0 and 1) or well-developed collaterals (Rentrop scores of 2 and 3). Collateral circulation scores were independently assessed by two experienced interventional cardiologists.

Cardiovascular magnetic resonance was performed with the Siemens 3.0-T Whole-body Scanner (MAGNETOM Verio, Tim System; Siemens Healthcare, Erlangen, Germany, UK). The CMR protocol included cine images and late gadolinium enhancement (LGE) imaging. As described in our previous study (15), 8-mm sections with no intersection gaps were obtained in the short-axis plane (from the base to the apex) and the long-axis plane of the left ventricle (LV) to perform cine cardiac MR and LGE imaging. Postprocessing analyses were performed by using the Siemens–Argus software.

Two experienced radiologists performed a visual analysis of all the images based on the American Heart Association 17-segment model of LV. Discrepancies were resolved by a discussion involving the presence of another senior investigator. Segments 1, 2, 7, 8, 13, 14, and 17 were considered left anterior descending (LAD) artery territory. Segments 5, 6, 11, 12, and 16 were considered left circumflex (LCX) artery territory. Segments 3, 4, 9, 10, and 15 were considered right coronary artery (RCA) territory (16). Taking coronary dominance into account, the inferoseptal segments, inferior segments, and inferolateral segments could be reassigned (17). The wall motion and extent of LGE were graded. The wall motion of each myocardial segment was scored on a scale of 1–4: 1 = normal, 2 = hypokinesia, 3 = akinesia, and 4 = dyskinesia. The extent of segmental wall enhancement was graded on a five-point scale: 1 = no hyperenhancement, 2 = hyperenhancement of 1–25% of tissue, 3 = hyperenhancement of 26–50% of tissue, 4 = hyperenhancement of 51–75% of tissue, and 5 = hyperenhancement of 76–100% of tissue (18, 19). LGE > 50% was considered to be a transmural scar, LGE between 1 and 50% was considered to be an endocardial scar with viable myocardium, and no LGE was considered to be absent of myocardial scar. Viable myocardium was defined as one with either no LGE or 1–50% LGE. Patients with left ventricular ejection fraction (LVEF) <50% were defined as having left ventricular dysfunction.

Statistical analysis was performed by using the Statistical Package for the Social Sciences (SPSS) 26.0 software (IBM Corp., Armonk, NY, USA). Normally distributed data were expressed as mean ± SD. Non-normally distributed data were expressed as median (interquartile range). Continuous variables were compared by using the t-test or the Mann–Whitney U test and categorical variables were compared by using the χ2 test or the Spearman's rank correlation test. Interobserver and intraobserver agreement were tested by using the Cohen's kappa. The correlation between wall motion score and extent of LGE was evaluated by using the Spearman's rank correlation test. A two-tailed p < 0.05 was considered to be statistically significant.

Baseline clinical and angiographic characteristics are presented in Table 1 and are ordered according to myocardial scar formation in CTO territories. Statistical analyses demonstrated that patients with myocardial scar were likely to be males and smokers and were more likely to have hypertension. The prevalence of myocardial scar in CTO territory detected by CMR (66%) was strikingly higher than traditional clinical evidence that would suggest, as only 11% of included study patients showed pathological Q waves on ECG. Angiographically, 79 patients (41%) showed at least one concomitant main artery with severe stenosis.

Of the 194 patients included in this study, a total of 128 patients (66%) had myocardial scar in CTO territory detected by CMR, while 66 patients (34%) did not have myocardial scar in CTO territory detected by CMR.

Late gadolinium enhancement imaging demonstrated myocardial scar in 417 segments (37%) out of 1,112 total segments. Over 50% LGE was observed in 198 (18%) CTO-related segments, while no scar was observed in 695 (63%) CTO-related segments (see Table 2). A total of 780 segments (70%) in CTO territory showed normal wall motion (score = 1). The Spearman's rank correlation test showed a significantly positive correlation between wall motion scores and LGE scores (r = 0.488, p < 0.001).

Left ventricular dysfunction was observed in 67 enrolled patients (35%) with an average LVEF of 36.4 ± 9.4%. Over 50% LGE was observed in 27% (103/381) of segments in CTO territory of patients with reduced LVEF (<50%), but only in 13% (95/731) of segments in CTO territory of those patients with normal systolic function (≥50%, p < 0.001). Patients with viable myocardia (LGE 0–50%) had higher LVEF (56.7 ± 13.5% vs. 48.3 ± 15.4%, p < 0.001, Figure 2A) and lower LV end-diastolic volume (LVEDV) (97.3 ± 33.8 ml vs. 121.7 ± 54.4 ml, p = 0.003, Figure 2B) and LV end-systolic volume (LVESV) (43.3 ± 24.4 ml vs. 68.6 ± 48.1 ml, p < 0.001, Figure 2C) than those with transmural scar.

Poorly-developed angiographic collaterals were observed in 30 patients (15%), while well-developed collaterals were found in 164 patients (85%). Interobserver and intraobserver agreement for the Rentrop grading of collaterals in 50 (26%) randomly selected patients was high (intraobserver agreement = 90%; the Cohen's kappa = 0.62, 95% CI: 0.32–0.92).

Of the patients with poorly-developed collaterals, a total of 21 (70%) patients had a myocardial scar in CTO territory, while 65% of patients with well-developed collaterals (107/164) showed scar in CTO territory. There was no significant correlation between the presence of myocardial scar and collateral status (p = 0.680). Nevertheless, 25% of segments subtended to CTO arteries showed over 50% LGE in patients with poorly-developed collaterals, while only 17% of segments subtended to CTO arteries showed over 50% LGE in those patients with well-developed collaterals (p = 0.010; Figure 3). Additionally, more viable CTO-related segments were found in the presence of well-developed collaterals than in the presence of poorly-developed collaterals, despite the non-statistically significant difference [591 out of 934 (63%) vs. 104 out of 178 (58%) segments, p = 0.237].

There were no significant differences between mean LVEF and ventricular volume in patients with poorly-developed collaterals when compared with those with well-developed collaterals (p-values: LVEF = 0.191, LVEDV = 0.360, LVESV = 0.300, Figures 2D–F). Typical case scenarios displaying patients with well- or poorly-developed collaterals and patients with different extents of myocardial scar are shown in Figure 4.

We reported a higher prevalence of myocardial scar detected by CMR than traditional clinical evidence in this study. However, this finding was consistent with previous work (20). We attribute this to the high sensitivity of CMR for recognizing endocardial scar, while the occurrence of myocardial infarction symptoms or Q waves on ECG relates more to transmural necrosis. Despite the considerable prevalence of myocardial scar, less than one-fifth of the segments showed transmural scar, suggesting that severe ischemic injury is limited in CTO territory. However, since the benefits of revascularization for CTO are still controversial, many researchers pay more attention to factors that predict better outcomes following CTO PCI. Studies have shown that myocardial viability is an important outcome predictor, as confined endocardial scar or absolute viability has been associated with functional recovery (21, 22). Schumacher et al. previously reported that 76% of patients had evidence of LGE and that only 5% of CTO segments demonstrated transmural scar tissue that could be detected by CMR (23). The notion that the ischemic myocardial scar is common but most limited within the endocardium would be consistent with our results. Particularly, an evident extensive myocardial scar was observed beyond CTO territories in a few patients, despite adjustment according to the dominance of coronary artery. Indeed, the complexity of vascular variability made it difficult to assign the 17 segments of LV to specific coronary arteries. As previously reported (17), the most specific segments including anterior, anteroseptal segments correspond to LAD, but no segments can be exclusively attributed to RCA or LCX. Additionally, Yajima et al. (24) also reported myocardial scar in territories adjacent or remote to CTO territories in a chronic myocardial infarction (MI) adult porcine model and explained it by endothelium swelling and microvasculature disruption. Therefore, the variable vascular territories and extensive microvasculature injury could account for the bias of myocardial viability analysis.

Ischemia is estimated to be responsible for around two-thirds of heart failure cases (25). In patients with ischemic ventricular dysfunction, the presence of CTO is associated with higher morbidity and poor prognosis (26). In this study, we reported that 35% of patients had impaired systolic functioning and that patients with transmural scar had worse functional statuses. A few observational studies investigating the outcomes of CTO PCI among patients with LV dysfunction have shown substantial LVEF improvement and decreased cardiac mortality, especially in those with severely impaired systolic functions (27–30). Interventional procedures are more challenging in these complicated patients. A study demonstrated that PCI was a safe strategy in patients with low LVEF ( ≤ 35%), as the angiographic success rate was high and similar to that in patients with LVEF > 35% without more periprocedural complications (31). However, these outcomes are still limited in non-randomized controlled trials (RCTs). The REVASC trial (A Randomized Trial to Assess Regional Left Ventricular Function After Stent Implantation in Chronic Total Occlusion) did not show any benefits related to CTO PCI. Therefore, experts recommend both the viability and functional assessment in patients with ischemic cardiomyopathy for CTO PCI (32).

There was no significant difference in the prevalence of myocardial scar or systolic functioning in patients with well- or poorly-developed collaterals and there were fewer non-viable segments in the CTO territory supplied by well-developed collaterals. The presence of well-developed collaterals may not directly predict myocardial viability, but could protect the myocardium from severe ischemic injury to some extent. In fact, whether collaterals could have protective effects on myocardial viability and contractility have long been a matter of debate. In a recent study, patients with well-developed collaterals showed less myocardial scar as assessed by quantitative CMR analysis and more retained systolic function in CTO territory (23). Previously, a few studies reported that collaterals had lower sensitivity for predicting myocardial viability (33, 34). The blood supply from collaterals to the totally occluded territory could be limited, especially when myocardial oxygen consumption increases. Additionally, myocardial viability and contractility could be affected by many concomitant factors including multiple-vessel stenosis and microvascular dysfunction. In the subgroup analysis of the EXPLORE (Evaluating Xience and Left Ventricular Function in Percutaneous Coronary Intervention on Occlusions After ST-Elevation Myocardial Infarction) trial, the presence of well-developed collaterals was shown to correlate with better outcomes. However, well-developed collaterals did not translate into better clinical outcomes. A recent meta-analysis showed that the presence of well-developed collaterals is not well-related to lower rates of acute MI (AMI) or mortality, but does increase the likelihood of successful CTO PCI (35, 36). These results indicated that well-developed collaterals should not be the only factor that affects prognosis.

There are several limitations inherent in this study. First, it was a single-center and observational study with a small sample size that enrolled patients with a range of LVEF—meaning the population with impaired LVEF accounted for a minority of patients. This does, however, reflect a real-world trend. Second, the LGE images were analyzed by using a semi-quantitative visual method rather than a quantitative method. However, the utility of quantitative CMR analysis by using commercial software is still limited and thresholds for signal intensity are not unified. Third, stress perfusion imaging was not performed on our patients, meaning that we lacked analysis of ischemic burden in CTO territory. Finally, collateral circulation was only assessed by using visualized angiography and functional intravascular evaluations were not employed.